activation functions added for bounded outputs #14
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
|
|
gabrieloks
force-pushed
the
feature/activation_function
branch
from
640b623 to
da2ffcc
Compare
mchantry
reviewed
Aug 8, 2024
gabrieloks
force-pushed
the
feature/activation_function
branch
from
8eaf33f to
373d67b
Compare
JesperDramsch
force-pushed
the
feature/activation_function
branch
from
373d67b to
85d45ae
Compare
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## develop #14 +/- ##
========================================
Coverage 99.83% 99.84%
========================================
Files 22 23 +1
Lines 1216 1277 +61
========================================
+ Hits 1214 1275 +61
Misses 2 2 ☔ View full report in Codecov by Sentry. |
mchantry
reviewed
Sep 10, 2024
mchantry
reviewed
Sep 11, 2024
|
Great work on this, thanks. |
for more information, see https://pre-commit.ci
for more information, see https://pre-commit.ci
mchantry
previously approved these changes
Sep 23, 2024
JesperDramsch
approved these changes
Sep 23, 2024
theissenhelen
pushed a commit
that referenced
this pull request
Sep 27, 2024
* activation functions added for bounded outputs * generalised fraction normalisation * precommit changes * precommit changes * chore: mv bounding to post-processors * feat: make bounding strategies torch Module * refactor: pre-build indices during init * refactor: nn.module in place modification * refactor: mv bounding to layers * refactor: implement bounding ModuleList * docs: add changelog * refactor: mv bounding config to models * feat: enable 1-1 variable remapping in preprocessors * test: create tests and fix code * test: add a test for hydra instantiating of bounding * fix: naming * refactor: reduce verboseness * docs: add comments * fix: inject name_to_index on initiation * fixed reading of statistics remapping * revert * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Update test_preprocessor_normalizer.py * Update encoder_processor_decoder.py * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * docs: added a comment on special keys * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * docs: add docstrings * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * fix: changelog --------- Co-authored-by: Jesper Dramsch <[email protected]>
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 PR is related to the open issue #13 and the aifs-mono PR https://github.com/ecmwf-lab/aifs-mono/pull/224.
The basic idea behind this PR is to implement bounding strategies for specific variables, chosen by the user, via the usage of activation functions. One evident example the total precipitation accumulation variable, which is not currently bounded, therefore prone to negative value predictions.
Closes #13