Update context order handling in ContextualDataset #2205
Closed
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
the
CLA Signed
|
This pull request was exported from Phabricator. Differential Revision: D52452236 |
Summary: X-link: facebook/Ax#2198 Currently ContextualDataset will re-order outputs to match the specified order of context_buckets. This has produced a somewhat confusing state, where at different places (ContextualDataset, the model, the surrogates) we have been passing around outcome lists or context lists that are expected to determine ordering in some way, but without a clear place where re-ordering is happening or where that order is being defined. D51906856 was a step towards clarifying this, by setting the policy that the modelbridge will be responsible for defining the desired ordering; it will then order the datasets accordingly; and everything downstream is expected to return the outputs in the same order as the datasets. That is, it is expected to return the outputs in the same order as the inputs, which seems the cleanest and most logical policy. This diff updates ContextualDataset in line with that philosophy so that hte output order is always the order that the outcomes are provided in the datasets. If the user wants a particular output order, that can be handled by ordering the input datasets accordingly rather than by specifying context buckets. With this change, the context_buckets input is no longer necessary since the bucket names can be read from the parameter decomposition, and the outcome order is taken from datasets. Reviewed By: saitcakmak Differential Revision: D52452236
bletham
added a commit
to bletham/Ax
that referenced
this pull request
Feb 14, 2024
Summary: X-link: pytorch/botorch#2205 Currently ContextualDataset will re-order outputs to match the specified order of context_buckets. This has produced a somewhat confusing state, where at different places (ContextualDataset, the model, the surrogates) we have been passing around outcome lists or context lists that are expected to determine ordering in some way, but without a clear place where re-ordering is happening or where that order is being defined. D51906856 was a step towards clarifying this, by setting the policy that the modelbridge will be responsible for defining the desired ordering; it will then order the datasets accordingly; and everything downstream is expected to return the outputs in the same order as the datasets. That is, it is expected to return the outputs in the same order as the inputs, which seems the cleanest and most logical policy. This diff updates ContextualDataset in line with that philosophy so that hte output order is always the order that the outcomes are provided in the datasets. If the user wants a particular output order, that can be handled by ordering the input datasets accordingly rather than by specifying context buckets. With this change, the context_buckets input is no longer necessary since the bucket names can be read from the parameter decomposition, and the outcome order is taken from datasets. Reviewed By: saitcakmak Differential Revision: D52452236
bletham
force-pushed
the
export-D52452236
branch
from
177a909
to
a3ef77a
Compare
|
This pull request was exported from Phabricator. Differential Revision: D52452236 |
bletham
added a commit
to bletham/Ax
that referenced
this pull request
Feb 16, 2024
Summary: X-link: pytorch/botorch#2205 Currently ContextualDataset will re-order outputs to match the specified order of context_buckets. This has produced a somewhat confusing state, where at different places (ContextualDataset, the model, the surrogates) we have been passing around outcome lists or context lists that are expected to determine ordering in some way, but without a clear place where re-ordering is happening or where that order is being defined. D51906856 was a step towards clarifying this, by setting the policy that the modelbridge will be responsible for defining the desired ordering; it will then order the datasets accordingly; and everything downstream is expected to return the outputs in the same order as the datasets. That is, it is expected to return the outputs in the same order as the inputs, which seems the cleanest and most logical policy. This diff updates ContextualDataset in line with that philosophy so that hte output order is always the order that the outcomes are provided in the datasets. If the user wants a particular output order, that can be handled by ordering the input datasets accordingly rather than by specifying context buckets. With this change, the context_buckets input is no longer necessary since the bucket names can be read from the parameter decomposition, and the outcome order is taken from datasets. Reviewed By: saitcakmak Differential Revision: D52452236
bletham
added a commit
to bletham/botorch
that referenced
this pull request
Feb 16, 2024
Summary: X-link: facebook/Ax#2198 Currently ContextualDataset will re-order outputs to match the specified order of context_buckets. This has produced a somewhat confusing state, where at different places (ContextualDataset, the model, the surrogates) we have been passing around outcome lists or context lists that are expected to determine ordering in some way, but without a clear place where re-ordering is happening or where that order is being defined. D51906856 was a step towards clarifying this, by setting the policy that the modelbridge will be responsible for defining the desired ordering; it will then order the datasets accordingly; and everything downstream is expected to return the outputs in the same order as the datasets. That is, it is expected to return the outputs in the same order as the inputs, which seems the cleanest and most logical policy. This diff updates ContextualDataset in line with that philosophy so that hte output order is always the order that the outcomes are provided in the datasets. If the user wants a particular output order, that can be handled by ordering the input datasets accordingly rather than by specifying context buckets. With this change, the context_buckets input is no longer necessary since the bucket names can be read from the parameter decomposition, and the outcome order is taken from datasets. Reviewed By: saitcakmak Differential Revision: D52452236
bletham
added a commit
to bletham/botorch
that referenced
this pull request
Feb 16, 2024
Summary: X-link: facebook/Ax#2198 Currently ContextualDataset will re-order outputs to match the specified order of context_buckets. This has produced a somewhat confusing state, where at different places (ContextualDataset, the model, the surrogates) we have been passing around outcome lists or context lists that are expected to determine ordering in some way, but without a clear place where re-ordering is happening or where that order is being defined. D51906856 was a step towards clarifying this, by setting the policy that the modelbridge will be responsible for defining the desired ordering; it will then order the datasets accordingly; and everything downstream is expected to return the outputs in the same order as the datasets. That is, it is expected to return the outputs in the same order as the inputs, which seems the cleanest and most logical policy. This diff updates ContextualDataset in line with that philosophy so that hte output order is always the order that the outcomes are provided in the datasets. If the user wants a particular output order, that can be handled by ordering the input datasets accordingly rather than by specifying context buckets. With this change, the context_buckets input is no longer necessary since the bucket names can be read from the parameter decomposition, and the outcome order is taken from datasets. Reviewed By: saitcakmak Differential Revision: D52452236
|
This pull request has been merged in e6d5469. |
facebook-github-bot
pushed a commit
to facebook/Ax
that referenced
this pull request
Feb 16, 2024
Summary: X-link: pytorch/botorch#2205 Pull Request resolved: #2198 Currently ContextualDataset will re-order outputs to match the specified order of context_buckets. This has produced a somewhat confusing state, where at different places (ContextualDataset, the model, the surrogates) we have been passing around outcome lists or context lists that are expected to determine ordering in some way, but without a clear place where re-ordering is happening or where that order is being defined. D51906856 was a step towards clarifying this, by setting the policy that the modelbridge will be responsible for defining the desired ordering; it will then order the datasets accordingly; and everything downstream is expected to return the outputs in the same order as the datasets. That is, it is expected to return the outputs in the same order as the inputs, which seems the cleanest and most logical policy. This diff updates ContextualDataset in line with that philosophy so that hte output order is always the order that the outcomes are provided in the datasets. If the user wants a particular output order, that can be handled by ordering the input datasets accordingly rather than by specifying context buckets. With this change, the context_buckets input is no longer necessary since the bucket names can be read from the parameter decomposition, and the outcome order is taken from datasets. Reviewed By: saitcakmak Differential Revision: D52452236 fbshipit-source-id: 499bd6772eb11e550598151389abe21716370f2b
stefanpricopie
pushed a commit
to stefanpricopie/botorch
that referenced
this pull request
Feb 27, 2024
Summary: Pull Request resolved: pytorch#2205 X-link: facebook/Ax#2198 Currently ContextualDataset will re-order outputs to match the specified order of context_buckets. This has produced a somewhat confusing state, where at different places (ContextualDataset, the model, the surrogates) we have been passing around outcome lists or context lists that are expected to determine ordering in some way, but without a clear place where re-ordering is happening or where that order is being defined. D51906856 was a step towards clarifying this, by setting the policy that the modelbridge will be responsible for defining the desired ordering; it will then order the datasets accordingly; and everything downstream is expected to return the outputs in the same order as the datasets. That is, it is expected to return the outputs in the same order as the inputs, which seems the cleanest and most logical policy. This diff updates ContextualDataset in line with that philosophy so that hte output order is always the order that the outcomes are provided in the datasets. If the user wants a particular output order, that can be handled by ordering the input datasets accordingly rather than by specifying context buckets. With this change, the context_buckets input is no longer necessary since the bucket names can be read from the parameter decomposition, and the outcome order is taken from datasets. Reviewed By: saitcakmak Differential Revision: D52452236 fbshipit-source-id: 499bd6772eb11e550598151389abe21716370f2b
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:
Currently ContextualDataset will re-order outputs to match the specified order of context_buckets.
This has produced a somewhat confusing state, where at different places (ContextualDataset, the model, the surrogates) we have been passing around outcome lists or context lists that are expected to determine ordering in some way, but without a clear place where re-ordering is happening or where that order is being defined.
D51906856 was a step towards clarifying this, by setting the policy that the modelbridge will be responsible for defining the desired ordering; it will then order the datasets accordingly; and everything downstream is expected to return the outputs in the same order as the datasets. That is, it is expected to return the outputs in the same order as the inputs, which seems the cleanest and most logical policy.
This diff updates ContextualDataset in line with that philosophy so that hte output order is always the order that the outcomes are provided in the datasets. If the user wants a particular output order, that can be handled by ordering the input datasets accordingly rather than by specifying context buckets. With this change, the context_buckets input is no longer necessary since the bucket names can be read from the parameter decomposition, and the outcome order is taken from datasets.
Reviewed By: saitcakmak
Differential Revision: D52452236