Deep Learning in Containers

This is a personal container-based Deep Learning setup I am using for development.

If you find this useful feel free to leave a ⭐

References

1. Deep learning with containers. Part 1 by Alexander Visheratin in TDS

What does it include?

• PyTorch 1.13 (GPU)
• Tensorflow 2.11 (CPU)
• Huggingface Transformers 4.25
• Several other useful ML/DL packages
• Nvidia GPU Driver 525.x.x
• CUDA 11.3.0
• cuDNN 8
• Multi-stage builds to reduce build times after adding new packages.
• Tensorboard external container for logging
• Separate CPU image

Note: DOES NOT include tensorflow for GPU as of now.

Getting started

Prerequisites

The docker images build here only work if you have an Nvidia GPU.

You should have the below mentioned tools set up and ready to go. I have linked the official installation instructions.

1. Docker
2. Docker Compose
3. Nvidia Container Toolkit

Getting it up and running

1. Clone the repo

2. navigate into the repo's root directory.

3. Use Docker Compose to build the images and launch the containers.

   $ sudo docker-compose up -d

   You can also use --build flag to specify you want to build/rebuild the image.

   $ sudo docker-compose up -d --build

You can also pull the prebuilt deep learning image from docker hub using this command. The badge at the top of this README.md file shows the latest version. For a specific version, eg: v0.1.3

$ sudo docker pull abhinand5/deeplearning-dev:v0.1.3

If you want to build the cpu image,

$ sudo docker-compose -f cpu.docker-compose.yaml up -d --build

To pull a CPU image you can use the tag abhinand5/deeplearning-dev:${versionName}-cpu, for example

$ sudo docker pull abhinand5/deeplearning-dev:latest-cpu

Note: Image size is bigger than whats shown in DockerHub, ~8GB after building on my machine for GPU and ~4.5GB for CPU-only.

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

License

MIT