-
Notifications
You must be signed in to change notification settings - Fork 415
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
BoxDecomposition cleanup #1490
BoxDecomposition cleanup #1490
Conversation
This pull request was exported from Phabricator. Differential Revision: D41172490 |
Codecov Report
@@ Coverage Diff @@
## main #1490 +/- ##
=========================================
Coverage 100.00% 100.00%
=========================================
Files 134 134
Lines 12402 12407 +5
=========================================
+ Hits 12402 12407 +5
📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more |
This pull request was exported from Phabricator. Differential Revision: D41172490 |
Summary: Pull Request resolved: pytorch#1490 - Change `compute_hypervolume` so that each BoxDecomposition subclass uses shared logic for the no-data case - [debatable] When `Y` is `None`, functions of Y like `box_decomp._neg_Y` are `None` rather than being unset attributes, so we do "if self._neg_Y is None" rather than catching an AttributeError. This makes catching type errors easier since otherwise Pyre is unhappy about references to the potentially-uninitialized attribute. - Took out unnecessary "register_buffer" calls (this happens automatically with `torch.nn.Module.setattr`) Differential Revision: D41172490 fbshipit-source-id: 01e3e159846b5b6cbd77763bdec00ee5d86aa50d
2bfc551
to
95fe8bc
Compare
Summary: Pull Request resolved: pytorch#1490 - Change `compute_hypervolume` so that each BoxDecomposition subclass uses shared logic for the no-data case - [debatable] When `Y` is `None`, functions of Y like `box_decomp._neg_Y` are `None` rather than being unset attributes, so we do "if self._neg_Y is None" rather than catching an AttributeError. This makes catching type errors easier since otherwise Pyre is unhappy about references to the potentially-uninitialized attribute. - Took out unnecessary "register_buffer" calls (this happens automatically with `torch.nn.Module.setattr`) Differential Revision: D41172490 fbshipit-source-id: 356485c84b2d6bc85b20ed34b7247ba1478057d2
95fe8bc
to
17cc28a
Compare
This pull request was exported from Phabricator. Differential Revision: D41172490 |
Summary: Pull Request resolved: pytorch#1490 - Change `compute_hypervolume` so that each BoxDecomposition subclass uses shared logic for the no-data case - [debatable] When `Y` is `None`, functions of Y like `box_decomp._neg_Y` are `None` rather than being unset attributes, so we do "if self._neg_Y is None" rather than catching an AttributeError. This makes catching type errors easier since otherwise Pyre is unhappy about references to the potentially-uninitialized attribute. - Took out unnecessary "register_buffer" calls (this happens automatically with `torch.nn.Module.setattr`) Differential Revision: D41172490 fbshipit-source-id: c6e26914be9b25ba268de3355a1ddc8ee9fb66d0
17cc28a
to
b8f5d26
Compare
This pull request was exported from Phabricator. Differential Revision: D41172490 |
Summary:
compute_hypervolume
so that each BoxDecomposition subclass uses shared logic for the no-data caseY
isNone
, functions of Y likebox_decomp._neg_Y
areNone
rather than being unset attributes, so we do "if self._neg_Y is None" rather than catching an AttributeError. This makes catching type errors easier since otherwise Pyre is unhappy about references to the potentially-uninitialized attribute.torch.nn.Module.setattr
)Differential Revision: D41172490