-
Notifications
You must be signed in to change notification settings - Fork 386
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
[FIX] Unify API #1023
Merged
Merged
[FIX] Unify API #1023
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
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
AzulGarza
reviewed
Nov 4, 2024
Todo:
|
cchallu
approved these changes
Feb 24, 2025
This was referenced Feb 24, 2025
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Outdated suggestions cannot be applied.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.
This is a large refactoring PR and open for discussion. The main goal of the PR is to unify API across different model types, and unify loss functions across different loss types.
Refactoring:
BaseWindows
,BaseMultivariate
andBaseRecurrent
intoBaseModel
, removing the need for separate classes and unifying model API across different model types. Instead, this PR introduces two model attributes, yielding four possible model options:RECURRENT
(True
/False
) andMULTIVARIATE
(True
/False
). We currently have a model for every combination except a recurrent multivariate model (e.g. a multivariate LSTM), however this is now relatively simple to add. In addition, this change allows to have models that can be recurrent or not, or multivariate or not on-the-fly, based on users' input. This also allows for easier modelling going forward.domain_map
functions.loss.domain_map
outside of models toBaseModel
TSMixer
,TSMixerx
andRMoK
tocommon.modules
Features:
DistributionLoss
now supports the use ofquantiles
inpredict
, allowing for easy quantile retrieval for all DistributionLosses.GMM
,PMM
andNBMM
) now support learned weights for weighted mixture distribution outputs.quantiles
inpredict
, allowing for easy quantile retrieval.ISQF
by adding softplus protection around some parameters instead of using.abs
Bug fixes:
MASE
loss now works.StudentT
increase default DoF to 3 to reduce unbound variance issues.eval: false
on the examples whilst not having any other tests, causing most models to effectively not being tested at all.Breaking changes:
input_size
to be given.TCN
andDRNN
are now windows models, not recurrent models.Tests:
common._model_checks.py
that includes a model testing function. This function runs on every separate model, ensuring that every model is tested on push.Todo: