Skip to content

Comment on ONNX runtime: Cross-platform accelerated machine learning

Comments

Would it not be better to use https://github.com/tinygrad/tinygrad as an intermediary framework?

Tinygrad is python only right? Can it provide gradients during C++ runtime as well? ONNX runtime have multiple language backends for inference.

One option to your case is OpenVino. It's written in C++ and has Python Bindings. Also, it can be used to train new nets. You can use ONNX files with OpenVino too.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.