Tutorial notebooks for hls4ml
There are several ways to run the tutorial notebooks:
The Python environment used for the tutorials is specified in the environment.yml
file. It can be setup like:
conda env create -f environment.yml
conda activate hls4ml-tutorial
Follow these steps to build a Docker image that can be used locally, or on a JupyterHub instance, e.g. as the single user image.
The Dockerfile was modified from one created using repo2docker
.
You can build the image straight from the Github:
docker build --build-arg NB_USER=jovyan --build-arg NB_UID=1000 https://github.com/hls-fpga-machine-learning/hls4ml-tutorial.git -f docker/Dockerfile
Alternatively, you can clone the repository and build locally (you might want to do this to add Vivado installation, for example):
git clone https://github.com/hls-fpga-machine-learning/hls4ml-tutorial.git
cd hls4ml-tutorial
# modify something
docker build --build-arg NB_USER=jovyan --build-arg NB_UID=1000 . -f docker/Dockerfile
Then to start the container:
docker run -p 8888:8888 <IMAGE ID>
When the container starts, the Jupyter notebook server is started, and the link to open it in your browser is printed.
We have prepared a set of slides with some introduction and more details on each of the exercises. Please find them here.