Feature add activation to BlockRNN (#2492) #2504
Merged
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
* Added support for specifying PyTorch activation functions (`ReLU`, `Sigmoid`, `Tanh`, or `None`) in the `BlockRNNModel`. * Ensured that activation functions are applied between fully connected layers, but not as the final layer. * Implemented a check to raise an error if an activation function is set but the model only contains one linear layer. * Updated documentation to reflect the new activation parameter and usage examples. * Added test cases to verify the correct application of activation functions and to handle edge cases.
Codecov ReportAttention: Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #2504 +/- ##
==========================================
- Coverage 93.79% 93.75% -0.05%
==========================================
Files 139 139
Lines 14741 14738 -3
==========================================
- Hits 13827 13818 -9
- Misses 914 920 +6 ☔ View full report in Codecov by Sentry. |
|
Should a warning be raised if |
madtoinou
requested changes
Aug 28, 2024
Co-authored-by: madtoinou <32447896+madtoinou@users.noreply.github.com>
Co-authored-by: madtoinou <32447896+madtoinou@users.noreply.github.com>
* Add a check that raise an error when activation is None and hidden_fc_sizes is greater than 0
dennisbader
requested changes
Aug 30, 2024
There was a problem hiding this comment.
Hi @SzymonCogiel and thanks for this great PR 🚀 We're close, I just had a couple more suggestions. Mainly that I would set a default activation, to make it easier for the user in case he uses multiple fc layers.
| @@ -430,11 +449,40 @@ def encode_year(idx): | |||
| logger=logger, | |||
| ) | |||
|
|
|||
| raise_if( | |||
There was a problem hiding this comment.
With the comment from above, we could remove this check
| ), | ||
| logger=logger, | ||
| ) | ||
| raise_if( |
There was a problem hiding this comment.
and we could remove this as well
Co-authored-by: Dennis Bader <dennis.bader@gmx.ch>
* _check_ckpt_parameters * Remove redundant raise_if
dennisbader
requested changes
Sep 5, 2024
There was a problem hiding this comment.
Thanks for the updates @SzymonCogiel. Last little adaptions and then we're ready to merge :)
Co-authored-by: Dennis Bader <dennis.bader@gmx.ch>
* Revert docstring _BlockRNNModule
madtoinou
approved these changes
Sep 9, 2024
dennisbader
approved these changes
Sep 13, 2024
eschibli
pushed a commit
to eschibli/darts
that referenced
this pull request
Oct 8, 2024
* Feature add activation to BlockRNN (unit8co#2492) * Added support for specifying PyTorch activation functions (`ReLU`, `Sigmoid`, `Tanh`, or `None`) in the `BlockRNNModel`. * Ensured that activation functions are applied between fully connected layers, but not as the final layer. * Implemented a check to raise an error if an activation function is set but the model only contains one linear layer. * Updated documentation to reflect the new activation parameter and usage examples. * Added test cases to verify the correct application of activation functions and to handle edge cases. * Update darts/models/forecasting/block_rnn_model.py Co-authored-by: madtoinou <32447896+madtoinou@users.noreply.github.com> * Update darts/models/forecasting/block_rnn_model.py Co-authored-by: madtoinou <32447896+madtoinou@users.noreply.github.com> * Feature add activation to BlockRNN (unit8co#2492) * Add a check that raise an error when activation is None and hidden_fc_sizes is greater than 0 * Update darts/models/forecasting/block_rnn_model.py Co-authored-by: Dennis Bader <dennis.bader@gmx.ch> * Feature add activation to BlockRNN (unit8co#2492) * _check_ckpt_parameters * Remove redundant raise_if * Update darts/models/forecasting/block_rnn_model.py Co-authored-by: Dennis Bader <dennis.bader@gmx.ch> * Feature add activation to BlockRNN (unit8co#2492) * Revert docstring _BlockRNNModule --------- Co-authored-by: madtoinou <32447896+madtoinou@users.noreply.github.com> Co-authored-by: Dennis Bader <dennis.bader@gmx.ch>
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.
Checklist before merging this PR:
Fixes #2492 .
Summary
ReLU,Sigmoid,Tanh, orNone) in theBlockRNNModel.Other Information
Consider using a logging warning instead of raising an error if an activation function is set but the model only contains one linear layer.