yinyang/ml repository overview

⁠Contents

• GPU-accelerated versions of:
  • PyTorch 1.7
  • Tensorflow 2.4
  • Onnx runtime 1.6
  • TensorRT 7.2
  • fast.ai 2.2
  • LightGBM 3.1
  • RAPIDS with BlazingSQL 0.18
  • Transformers 4.3
• Common Python packages for machine learning.
• Jupyter notebook IDE with multiple programming languages (inherited from yinyang/base⁠ image).

⁠How to install

⁠Step 1: install Nvidia CUDA.

• Linux: run the commands from Nvidia⁠.

• Windows:

  • Sign up for Windows Insider Program, and switch to the fast track.
  • Install Windows Subsystem for Linux (WSL), and then Ubuntu from Microsoft App Store⁠.
  • Install the Nvidia driver for WSL⁠.
  • Open WSL, run the commands from Nvidia⁠.

⁠Step 2: install Nvidia-Docker.

• Follow the instructions from Nvidia⁠.

• On Windows, do not use Docker Desktop for Windows. Instead, install nvidia-docker using WSL command line as above.

⁠Step 3: pull this image.

⁠How to run

⁠Step 1: run this docker image:

• Linux and Windows WSL: sudo docker run --rm -it --gpus all --net host --shm-size 1g yinyang/ml

• Jetson: sudo docker run --rm -it --gpus all --net host --shm-size 1g yinyang/ml:jetson

⁠Step 2: start Jupyter notebook inside docker container:

jupyter notebook password
jupyter notebook --ip 0.0.0.0 --no-browser --allow-root

⁠Step 3: open localhost:8888⁠ in a web browser.

⁠License

Creative Commons⁠