Change how ModelListGP.posterior evaluates sub-models #1854
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
facebook-github-bot
added
CLA Signed
|
This pull request was exported from Phabricator. Differential Revision: D46364187 |
saitcakmak
added a commit
to saitcakmak/botorch
that referenced
this pull request
Jun 1, 2023
…ytorch#1854) Summary: Pull Request resolved: pytorch#1854 Prior to this change, `ModelListGP.posterior` would call it's submodels directly and work with the returned MVNs, skipping any processing in the individual `posterior` methods. This was particularly consequential for the `MultiTaskGP` (and subclasses) where skipping the `posterior` meant changing the expected shape of the `X` required by the `posterior`. With this change, `ModelListGP.posterior` evaluates the `posterior` methods for each of its submodels, and combines the MVNs into a single MVN where applicable. For `MultiTaskGP`, this makes it so that the `posterior` can be evaluated using the same inputs regardless of whether the mdoel is wrapped in a `ModelListGP` or not. Differential Revision: D46364187 fbshipit-source-id: 8e0ee449070607224fa6cd8510ed973a0c0da0c7
saitcakmak
force-pushed
the
export-D46364187
branch
from
52528ee to
be5019b
Compare
|
This pull request was exported from Phabricator. Differential Revision: D46364187 |
saitcakmak
added a commit
to saitcakmak/botorch
that referenced
this pull request
Jun 3, 2023
…ytorch#1854) Summary: Pull Request resolved: pytorch#1854 Prior to this change, `ModelListGP.posterior` would call it's submodels directly and work with the returned MVNs, skipping any processing in the individual `posterior` methods. This was particularly consequential for the `MultiTaskGP` (and subclasses) where skipping the `posterior` meant changing the expected shape of the `X` required by the `posterior`. With this change, `ModelListGP.posterior` evaluates the `posterior` methods for each of its submodels, and combines the MVNs into a single MVN where applicable. For `MultiTaskGP`, this makes it so that the `posterior` can be evaluated using the same inputs regardless of whether the mdoel is wrapped in a `ModelListGP` or not. Differential Revision: D46364187 fbshipit-source-id: e3af2a582037c66dd1f24d565b789f2085c9a28c
saitcakmak
force-pushed
the
export-D46364187
branch
from
be5019b to
1b54268
Compare
|
This pull request was exported from Phabricator. Differential Revision: D46364187 |
saitcakmak
added a commit
to saitcakmak/botorch
that referenced
this pull request
Jun 5, 2023
…ytorch#1854) Summary: Pull Request resolved: pytorch#1854 Prior to this change, `ModelListGP.posterior` would call it's submodels directly and work with the returned MVNs, skipping any processing in the individual `posterior` methods. This was particularly consequential for the `MultiTaskGP` (and subclasses) where skipping the `posterior` meant changing the expected shape of the `X` required by the `posterior`. With this change, `ModelListGP.posterior` evaluates the `posterior` methods for each of its submodels, and combines the MVNs into a single MVN where applicable. For `MultiTaskGP`, this makes it so that the `posterior` can be evaluated using the same inputs regardless of whether the mdoel is wrapped in a `ModelListGP` or not. Differential Revision: D46364187 fbshipit-source-id: 0a506c0a98454e6cfa9ac0068602f8098f31010b
saitcakmak
force-pushed
the
export-D46364187
branch
from
1b54268 to
ba3d108
Compare
|
This pull request was exported from Phabricator. Differential Revision: D46364187 |
Codecov Report
@@ Coverage Diff @@
## main #1854 +/- ##
=========================================
Coverage 100.00% 100.00%
=========================================
Files 171 171
Lines 15055 15051 -4
=========================================
- Hits 15055 15051 -4
📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more |
saitcakmak
added a commit
to saitcakmak/botorch
that referenced
this pull request
Jun 6, 2023
Summary: Prior to this change, `MultiTaskGP` could be evaluated using inputs of shape `batch x d` when using the model directly, and using inputs of shape `batch x d + 1` when the model was wrapped in a `ModelListGP`. This diff (combined with pytorch#1854) legitimizes both input shapes regardless of what API is used to evaluate the model. This will unify the APIs and increase the flexibility in how the model is used. Differential Revision: D46496775 fbshipit-source-id: 3a11aa5a1a6b79456a57491dac561a3a1860c14b
This was referenced Jun 6, 2023
saitcakmak
added a commit
to saitcakmak/botorch
that referenced
this pull request
Jun 6, 2023
…h#1868) Summary: Pull Request resolved: pytorch#1868 Prior to this change, `MultiTaskGP` could be evaluated using inputs of shape `batch x d` when using the model directly, and using inputs of shape `batch x d + 1` when the model was wrapped in a `ModelListGP`. This diff (combined with pytorch#1854) legitimizes both input shapes regardless of what API is used to evaluate the model. This will unify the APIs and increase the flexibility in how the model is used. Reviewed By: Balandat Differential Revision: D46496775 fbshipit-source-id: f44f5e529d9105d203923013101c37a8c1697e52
…h#1868) Summary: Pull Request resolved: pytorch#1868 Prior to this change, `MultiTaskGP` could be evaluated using inputs of shape `batch x d` when using the model directly, and using inputs of shape `batch x d + 1` when the model was wrapped in a `ModelListGP`. This diff (combined with pytorch#1854) legitimizes both input shapes regardless of what API is used to evaluate the model. This will unify the APIs and increase the flexibility in how the model is used. Differential Revision: https://internalfb.com/D46496775 fbshipit-source-id: e79dee0cc0ba0cd03fb97382cc1eb9ef4e5a8371
Summary: Pull Request resolved: pytorch#1854 Prior to this change, `ModelListGP.posterior` would call it's submodels directly and work with the returned MVNs, skipping any processing in the individual `posterior` methods. This was particularly consequential for the `MultiTaskGP` (and subclasses) where skipping the `posterior` meant changing the expected shape of the `X` required by the `posterior`. With this change, `ModelListGP.posterior` evaluates the `posterior` methods for each of its submodels, and combines the MVNs into a single MVN where applicable. For `MultiTaskGP`, this makes it so that the `posterior` can be evaluated using the same inputs regardless of whether the model is wrapped in a `ModelListGP` or not. Differential Revision: D46364187 fbshipit-source-id: a2b8293f71fd4be28d7e665e8fc0ab338b9bf601
saitcakmak
force-pushed
the
export-D46364187
branch
from
ba3d108 to
a7a3491
Compare
|
This pull request was exported from Phabricator. Differential Revision: D46364187 |
saitcakmak
changed the title
facebook-github-bot
pushed a commit
that referenced
this pull request
Jun 7, 2023
Summary: Pull Request resolved: #1868 Prior to this change, `MultiTaskGP` could be evaluated using inputs of shape `batch x d` when using the model directly, and using inputs of shape `batch x d + 1` when the model was wrapped in a `ModelListGP`. This diff (combined with #1854) legitimizes both input shapes regardless of what API is used to evaluate the model. This will unify the APIs and increase the flexibility in how the model is used. Reviewed By: Balandat Differential Revision: D46496775 fbshipit-source-id: 1de8e38ef1c4643f9006675058761e1b252acc03
|
This pull request has been merged in ffff366. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary:
Prior to this change,
ModelListGP.posteriorwould call it's submodels directly and work with the returned MVNs, skipping any processing in the individualposteriormethods. This was particularly consequential for theMultiTaskGP(and subclasses) where skipping theposteriormeant changing the expected shape of theXrequired by theposterior.With this change,
ModelListGP.posteriorevaluates theposteriormethods for each of its submodels, and combines the MVNs into a single MVN where applicable. ForMultiTaskGP, this makes it so that theposteriorcan be evaluated using the same inputs regardless of whether the mdoel is wrapped in aModelListGPor not.Differential Revision: D46364187