Comment by ComputerGuru

4 months ago

If you’re not writing/modifying the model itself but only training, fine tuning, and inferencing, ONNX now supports these with basically any backend execution provider without needing to get into dependency version hell.

0 comments