Conversation
|
Your PR no longer requires formatting changes. Thank you for your contribution! |
|
I split up the functions a bit more so we can reuse some more functionality in the ChainRules extension as well. Additionally, I cleaned up some of the index magic as preparatory work to more easily define these functions in TensorKit as well, hopefully removing a bit more of the boiler plate code that is required there as well. |
lkdvos
left a comment
lkdvos
left a comment
There was a problem hiding this comment.
I think I've addressed your comments, if tests pass this should be good to go from my end.
|
LGTM, they aren't charging us by the PR so we can do touchups if needed later |
|
TBH the cuda queue is super long rn and this doesn't affect the cutensor stuff, I'm happy to just be bad and bypass merge |
2 participants
Move this logic out of #243 so we can merge it without dealing with weird Enzyme segfaults.