-
Notifications
You must be signed in to change notification settings - Fork 313
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
Update get_sobol_mbm_generation_strategy
to allow specification of transforms
#3188
Open
ltiao
wants to merge
6
commits into
facebook:main
Choose a base branch
from
ltiao:export-D67357004
base: main
Could not load branches
Branch not found: {{ refName }}
Loading
Could not load tags
Nothing to show
Loading
Are you sure you want to change the base?
Some commits from the old base branch may be removed from the timeline,
and old review comments may become outdated.
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
…ationFeature (facebook#3023) Summary: Pull Request resolved: facebook#3023 **Context:** The values corresponding to map keys are propagated as part of the ObservationFeatures' `metadata` dict field. We require a way to place it in the `parameters` dict field so that it can be used later on. This generalized transform is able to take user-specified entries from an `ObservationFeatures`'s `metadata` field and place it within its `parameters` field, and update the search space accordingly to reflect this. This implements a new transform, `MetadataToFloat`, that extracts specified fields from each `ObservationFeature` instance's metadata and incorporates them as parameters. Furthermore, it updates the search space to include the specified field as a `RangeParameter` with bounds determined by observations provided during initialization. This process involves analyzing the metadata of each observation feature and identifying relevant fields that need to be included in the search space. The bounds for these fields are then determined based on the observations provided during initialization. Differential Revision: D65430943
… that provides sensible defaults for MapData (facebook#3155) Summary: Pull Request resolved: facebook#3155 This adds a specialization of the `MetadataToFloat` Transform, `MapKeyToFloat`, that provides sensible default settings to allow for intercepting map metric data appearing in the ObservationFeatures' metadata. Additionally, for the purposes of specifying `fixed_features` down the line, when `_transform_observation_feature` is given an empty `ObservationFeatures` (more specifically, an `ObservationFeatures` with an empty `parameters` dict), it will populate it with the *upper bound* associated with each metadata key. Differential Revision: D66945078
… rate for MapData (facebook#3106) Summary: Pull Request resolved: facebook#3106 This refines the logic for calculating per-metric subsampling rates in `MapData.subsample` and incorporates a (probably premature) performance optimization, achieved by utilizing binary search on a sorted list instead of linear search. Differential Revision: D66366076 Reviewed By: Balandat
…ta to Observations (facebook#3001) Summary: Pull Request resolved: facebook#3001 This updates `observations_from_data` to include progression info as observation feature metadata by default. More specifically: - Updates `observations_from_data` to subsume behavior of `observations_from_map_data` as special case. - Updates calls to `observations_from_map_data` to instead call `observations_from_data` - Removes `observations_from_map_data` which is used exclusively by `MapTorchModelBridge` Differential Revision: D65255312 Reviewed By: saitcakmak
…a, [Ax][WIP/Not Ready] Putting it all together (subclass TorchModelBridge) (facebook#3112) Summary: Pull Request resolved: facebook#3112 * Provide a new method `latest` for `MapData` to retrieve the *n* most recently observed values for each (arm, metric) group, determined by the `map_key` values, where higher implies more recent. * Update `observations_from_data` to optionally utilize `latest` and retain only the most recently observed *n* values (the new option, if specified alongside the existing subsampling options, will now take precedence). * Modify the "upcast" `MapData.df` property to leverage `latest`, which is a special case with *n=1*. * Revise the docstring to reflect changes in the pertinent methods, as well as update related methods like `subsample` to ensure uniform and consistent writing. Differential Revision: D66434621
…transforms Summary: Allows the user to customize the list of transforms to use for the Sobol-MBM benchmarking method. Differential Revision: D67357004
facebook-github-bot
added
the
CLA Signed
Do not delete this pull request or issue due to inactivity.
label
Dec 17, 2024
This pull request was exported from Phabricator. Differential Revision: D67357004 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Labels
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: Allows the user to customize the list of transforms to use for the Sobol-MBM benchmarking method.
Differential Revision: D67357004