Why are Neural SDEs trained with TrackerAdjoint()?

Specific Domains Machine Learning
, ,
1

I thought that Tracker was mostly replaced by Zygote and Enzyme for Reverse-mode Autodiff in the sciml ecosystem.
Why is it still being used for computing the sensitivities in DiffeqFlux?

2

Just limitations with higher order AD in Zygote, though those issues were actually fixed a few years ago so it could probably update its default. Also some of @avikpal 's Lux changes for higher order autodiff might come into play here too. So we probably just need to reassess it.

1 Like