Repository navigation
fix(stubs): add torch.half dtype alias - #5365
mangeshraut712 wants to merge 2 commits into
Conversation
|
This pull request has been imported. If you are a Meta employee, you can view this in D127482167. (Because this pull request was imported automatically, there will not be any future comments.) |
stroxler
left a comment
There was a problem hiding this comment.
Review automatically exported from Phabricator review in Meta.
|
This pull request has been merged in b0feebb. |
|
Hey — CodeSync landed this on main with me as the commit author (thanks). GitHub still shows the PR as closed not merged ( Could you mark CodeSync landings as merged on the GitHub side (or reopen + merge via GitHub) so external contributors get the PR merge credit? Same issue on #5365 #5373 #5367 #5369. |
|
Hi Pyrefly team, Thank you for reviewing and landing the changes from my account, including #5365 and the related stub/fix PRs. The commits on main correctly credit mangeshraut712 as author. I noticed that CodeSync closes the original PRs and adds the “Merged” label, but their GitHub merged status remains false. They therefore don’t appear in is:merged searches or counts based on GitHub-merged PRs. An accurate public contribution record matters to me for portfolio and professional visibility. I’m not asking for a duplicate merge or any rewrite of history, only confirmation of the expected workflow and the recommended way for external contributors to reference these accepted changes. Is closed-but-not-merged the normal outcome of CodeSync? If so, should I cite the landing commit and “Merged” label, or is there another preferred way? Disclosure: I used AI assistance to draft this message. The submitted changes themselves were also produced with bot assistance. Thanks for accepting the work and for any guidance. |
Summary
torch.halfwas missing from the tensor-shapes Torch stubs even though it is a public alias oftorch.float16. Add the alias beside the existing dtype aliases and a focused regression that checks runtime identity anddtype=use.Fixes #5361
Test plan
test_dtype_aliases.pyfortorch.half