Dan Foreman-Mackey
Dan Foreman-Mackey
Thanks for reporting this! I'm not totally surprised that things don't go so well if we nest JAX functions inside of a `pure_callback`, but I agree that hanging isn't a...
> Unfortunately, I have found performance to be quite different (also when using `scipy.linalg.eig`). Interesting! Can you say more about what performance differences you're seeing? I believe that the JAX...
This was fixed in https://github.com/openxla/xla/pull/21990 with a test added in https://github.com/jax-ml/jax/pull/26173. The approach here is to disable CPU asynchronous dispatch when running within the body of a callback. See the...
@dpshelio, @Cadair, @joyvelasquez — This is the review thread for the paper. All of our correspondence will happen here from now on. Thanks again for agreeing to participate! 👉 **Please...
@dpshelio, @Cadair — This is a ping to please get started with your reviews ASAP. Many thanks!
@joyvelasquez — I wanted to check in to see if you could update this thread with a summary of where things stand with respect to the recommended revisions. Thanks!!
@joyvelasquez — Any updates here? Let us know the status of your work addressing these comments.
@joyvelasquez — Thanks for the update - sounds good!
Thanks for the update @joyvelasquez! @dpshelio, @Cadair — Can you take another look in light of these changes? Thanks!
@dpshelio, @Cadair — A reminder to keep this on your radar. Thanks!