Adam Coogan

Results 3 comments of Adam Coogan

This is a problem for me as well. Any ideas about how to deal with this?

I've been trying to experiment with different initialization schemes for MLPs and came across this issue. Is there a simple way to use `tree_at` to apply e.g. [Lecun initialization](https://www.tensorflow.org/api_docs/python/tf/keras/initializers/LecunNormal) to...

Bumping this -- it would be quite helpful for me as well!