Installation¶
With Python >= 3.10:
This pulls in everything you need to build and train sparse models: TensorFlow, Keras 3, HGQ2 (the quantization backend), and our built-in monitoring and plotting tools.
Converting to HLS¶
Deploying a trained model to FPGA firmware additionally needs hls4ml. Support for sparsepixels is in an open pull request that is not yet merged, so for now install hls4ml from the PR branch:
See HLS Conversion for details.