Acknowledgements

Special Thanks

✨ @andife for a thoughtful series of contributions strengthening export provenance, GitHub Actions security, and dependency maintenance.

✨ @clementpoiret for initiating Equinox support, later returning to help harden the integration, and for Equimo, which brings modern vision models—such as DINOv3—to JAX/Equinox.

✨ @justinchuby for introducing onnx-ir as a scalable and more efficient way to handle ONNX model construction.

✨ @atveit for introducing us to gpt-oss-jax-vs-torch-numerical-comparison.

✨ @benmacadam64 for championing the complex-number handling initiative.

✨ @Artoriuz for the original depth_to_space contribution and guidance on the associated NCHW/NHWC export path.

✨ @lutzroeder for making shapes internal to ONNX function visible in his great Netron viewer. - ONNX: Function value_info support #1447

✨ tumaer/JAXFLUIDS for contributing valuable insights rooted in physics simulation use cases.

✨ @limarta for the elegant jaxpr-to-ONNX demonstration that inspired this project.

Contributors

Example contributions: @burakssen, @Cadynum, @clementpoiret, @PVirie, @Artoriuz

Plugin contributions: @burakssen, @clementpoiret, @Clouder0, @rakadam, @benmacadam64, @jakelee8

Community

Thanks to the community members involved in:

• Flax Feature Request #4430
• JAX Feature Request #26430