A package for machine learning inference in FPGAs. We create firmware implementations of machine learning algorithms using high level synthesis language (HLS). We translate traditional open-source machine learning package models into HLS that can be configured for your use-case!
Master : https://github.com/hls-fpga-machine-learning/hls4ml
These instructions will get you a copy of the project up and running on your local machine for development and testing purposes. See deployment for notes on how to deploy the project on a live system.
Quartus Prime license: the package only supports Intel devices, a license is required for the simulation and synthesis of HLS code.
- Tested on version 19.1
In addition the following dependencies are required:
Python 3
Change directory to /keras-to-hls
python keras-to-hls.py -c keras-config.yml
Configuration options for the HLS translation of Keras models.
KerasJson, KerasH5: For Keras translation, you are required to provide json and h5 model files.
Examples are in the directory: example-keras-model-files
OutputDir: Directory where your HLS project will go
DefaultPrecision: This is the default type of the weights, biases, accumulators, input and output vectors. This can then be further modified by the firmware/parameters.h file generated in your HLS project.
Note: The package optimizes for minimum latency. The Reuse factor support is being finalized will allow a better latency / area tradeoff.
cd my-hls-test
make myproject-fpga
- Hamza Javed - Intel Optimization - CERN
See also the original contributors who participated in this project.
- The amazing HLS4ML Team
- My supervisors Maurizio, Vladimir and Jennifer