Igor Kuszczak

Thank you for the quick response. You mentioned that setting `atol` to 10% approximates a concentrated force/ point load. What is the closest approximation one could get with this approach?...

Hi @patrick-kidger, I updated the original post with a condensed MWE. In the above code, when I use `optx.RecursiveCheckpointAdjoint()`, I am able to recover the correct gradients. However, when I...

Hi Patrick, Thank you for your thorough response. It is true that the solution to a minimisation problem is independent of the initial parameters and should lead to zero gradients....

Hi, Thanks for the quick response. I am currently trying to use optimistix to implement Model Agnostic Meta-Learning and its implicit version (https://arxiv.org/abs/1909.04630). I was considering using a `multi_step_solve` approach...

Thanks Patrick, I will try to make a PR on this :)

Hi @hawkinsp, Thank you for the quick response. I followed your instructions but encountered the same error. To provide more context, the error occurs during batched operations with a model...

@hawkinsp, here is the HLO dump for the run with 0.4.28 [dump.zip](https://github.com/google/jax/files/15435148/dump.zip)

Hi @hawkinsp, I am checking in to see if there are any updates on this issue. Any information would be helpful. Thanks!

Hi @hawkinsp, Apologies for the late response. I tried following the installation instructions for JAX nightly but I ended up with the following error: ``` [INFO 06-10 00:27:06] metatopia: Running...

I have now tried to do the same thing on Julia 1.4.1 and I am getting the following errors: ``` signal (6): Aborted │ in expression starting at /home/igork/.julia/packages/ADCME/94vEM/deps/build.jl:70 │...