Status and Features
Status
The latest version (built from main
) is 0.9.0.dev65+gafed23b1.
The stable version (released on PyPI) is 0.8.1.
See the Release Notes section for a changelog.
Features
A list of supported ML frameworks, HLS backends, and neural network architectures, including a summary table is below. Dependencies are given in the Setup page.
ML framework support:
(Q)Keras
PyTorch (limited)
(Q)ONNX (in development)
Neural network architectures:
Fully connected NN (multilayer perceptron, MLP)
Convolutional NN
Recurrent NN (LSTM)
Graph NN (GarNet)
HLS backends:
Vivado HLS
Intel HLS
Vitis HLS (experimental)
A summary of the on-going status of the hls4ml
tool is in the table below.
ML framework/HLS backend |
(Q)Keras |
PyTorch |
(Q)ONNX |
Vivado HLS |
Intel HLS |
Vitis HLS |
---|---|---|---|---|---|---|
MLP |
|
|
|
|
|
|
CNN |
|
|
|
|
|
|
RNN (LSTM) |
|
|
|
|
|
|
GNN (GarNet) |
|
|
|
|
|
|
Other feature notes:
hls4ml
is tested on Linux, and supportsVivado HLS versions 2018.2 to 2020.1
Intel HLS versions 20.1 to 21.4
Vitis HLS versions 2022.2 to 2024.1
Windows and macOS are not supported
BDT support has moved to the Conifer package
Example Models
We also provide and document several example hls4ml
models in this GitHub repository, which is included as a submodule.
You can check it out by doing git submodule update --init --recursive
from the top level directory of hls4ml
.