python
`tensorflow`: Add missing members to the `tensorflow.keras.layers` module.
![github-actions[bot] avatar](https://avatars.githubusercontent.com/in/15368?v=4 "Published by a pub.dev verified publisher"){kind=small}
@hmc-cs-mdrissi I've tried to add the MultiHeadAttention's __call__ to the stubtest_allowlist.txt, but it is still detected as an error. Did I misunderstand how to add functions to the stubtest allowlist ?
Edit: after spending a bunch of time trying to make CI pass without success, I've removed MultiHeadAttention from the PR.
@hmc-cs-mdrissi Quite a few things from your stubs are only present in tensorflow 2.15, would it be ok to bump the stubs' tensorflow version ? (in another PR, since it breaks a few things)
Quite a few things from your stubs are only present in tensorflow 2.15, would it be ok to bump the stubs' tensorflow version ?
Yes, we generally aim for the latest version. And agree it should be done in a separate PR :)
Yes feel free to bump the tf version in it's own pr. Stubtest will fail on new arguments and need some methods fixed. Usually filling in new arguments for stubtest is main work needed for upgrading version.
This has been a busy work week. I’ll take a pass over all your open prs that I haven’t already approved tomorrow.
@JelleZijlstra I don't understand why the CI is failing. It was passing after your merge commit, so I tried to revert the test commit I did earlier. This caused the CI to fail again, so I removed it, and now the CI is failing on the merge commit.
Is it some kind of caching issue ?
Some weird pytype issue, I'm unfortunately not sure how to address that. Seems like it also failed on the merge commit, though?