Only allow deserialization of KerasSaveables by module and name.
#21429
Conversation
This file contains hidden or 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
Codecov Report❌ Patch coverage is Additional details and impacted files@@ Coverage Diff @@
## master #21429 +/- ##
==========================================
- Coverage 82.72% 82.72% -0.01%
==========================================
Files 565 565
Lines 55222 55245 +23
Branches 8609 8612 +3
==========================================
+ Hits 45683 45700 +17
- Misses 7428 7433 +5
- Partials 2111 2112 +1
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
Arbitrary functions and classes are not allowed. - Made `Operation` extend `KerasSaveable`, this required moving imports to avoid circular imports - `Layer` no longer need to extend `KerasSaveable` directly - Made feature space `Cross` and `Feature` extend `KerasSaveable` - Also dissallow public function `enable_unsafe_deserialization`
hertschuh
force-pushed
the
serialization_fix
branch
from
fdf490e to
c69516b
Compare
google-ml-butler
bot
added
kokoro:force-run
ready to pull
Mohamed-Ashraf273
pushed a commit
to Mohamed-Ashraf273/keras
that referenced
this pull request
Jul 8, 2025
…eras-team#21429) Arbitrary functions and classes are not allowed. - Made `Operation` extend `KerasSaveable`, this required moving imports to avoid circular imports - `Layer` no longer need to extend `KerasSaveable` directly - Made feature space `Cross` and `Feature` extend `KerasSaveable` - Also dissallow public function `enable_unsafe_deserialization`
kharshith-k
pushed a commit
to kharshith-k/keras
that referenced
this pull request
Jul 23, 2025
…eras-team#21429) Arbitrary functions and classes are not allowed. - Made `Operation` extend `KerasSaveable`, this required moving imports to avoid circular imports - `Layer` no longer need to extend `KerasSaveable` directly - Made feature space `Cross` and `Feature` extend `KerasSaveable` - Also dissallow public function `enable_unsafe_deserialization`
hertschuh
added a commit
that referenced
this pull request
Nov 13, 2025
* added requirements-tensorflow-tpu.txt and tpu configuration in .kokoro * updated .kokoro/github/ubuntu/tpu/build.sh with jax and torch backend configs * Changed the tpu CI config files path to .github from .kokoro * Added new job in .github/workflows/actions.yml to run TPU tests * fixed runs-on option in acvtions.yml for tpu_build job to run on self hosted TPU based runner * Added another runner in the actions TPU job * Update continuous.cfg updated build file path * Update presubmit.cfg updated build file path * Update actions.yml Updated tpu_build job of actions.yml with specific runner label * Developed Dockerfile for TPU build job in actions.yml * Update actions.yml Added container section * Included few more runners in tpu_build job * Using linux-x86-ct6e-44-1tpu * Modified requirement-commmon.txt and updated requirements-tensorflow-tpu.txt * Added Dtypes_TPU_tests.py and requirements-jax-tpu.txt * Progress bar now handles `steps_per_execution`. (#21422) Progress bar would always report the starting batch + 1 at the end of the batch. Now it takes into account `steps_per_execution` for the last batch reported. Fixes #20861 * Fix symbolic call of `logsumexp` with int axis. (#21428) Using `keras.ops.math.logsumexp` with an int for `axis` in a functional model would throw an error. * Only allow deserialization of `KerasSaveable`s by module and name. (#21429) Arbitrary functions and classes are not allowed. - Made `Operation` extend `KerasSaveable`, this required moving imports to avoid circular imports - `Layer` no longer need to extend `KerasSaveable` directly - Made feature space `Cross` and `Feature` extend `KerasSaveable` - Also dissallow public function `enable_unsafe_deserialization` * commented tensorflow deps * Added log of dtypes_test_tpu.py and the test script for the same * modified dtypes_test_tpu.py as per pre-commit standards * Added TPU initiaization and teardown functionalities in conftest.py, developed dtypes_new_test.py to use requires_tpu marker * Added dtypes_test_TPU.py and dtypes_new_test.py, modified conftest.py * Added Dcokerfile and tests list command * Updated Dockerfile * Restored Dockerfile to previous changes * updated actions.yml file to install and configure docker engine on self hosted runner, build the image and check TPU support on jax backend * updated actions.yml file to include container option * updated actions.yml file to include container option without volume binding * updated actions.yml file to change TPU * Updated container path in build-and-test-on-tpu job * seperated TPU workflow from actions.yml * updated trigger condition for TPU tests workflow * updated container usage configuration for TPU tests workflow * updated env vars for TPU tests workflow * updated env vars parsing syntax in TPU tests workflow * updated env vars syntax in TPU tests workflow * updated env vars syntax in TPU tests workflow * updated env vars syntax in TPU tests workflow * updated env vars syntax in TPU tests workflow * updated image name in TPU tests workflow * updated image name with generic ubuntu image * updated tpu-tests to use ghcr * updated tpu-tests to store built image as local tar * updated image name from ubuntu:22.04 to docker:24.0-cli in tpu tests workflow * updated image name from docker:24.0-cli to ubuntu:22.04 in tpu tests workflow and added a step to install docker client * added volume mount from host in load-and-test-job * Reverted tpu-tests.yml to version using ghcr.io for image storage * Removed custom dtypes_test files for TPU testing and restored original actions.yml * Updated tpu-tests.yml to pull image from GCP artifact registry * Resolved conflicts in actions.yml * Added a workflow to check service accounts associated with self hosted runners * Made find_sa.yml specific to linux-x86-ct6e-44-1tpu * Added container tag to find_sa.yml * Checking SA for linux-x86-ct5lp-112-4tpu * Checking SA for linux-x86-ct6e-44-1tpu-nxgm7-runner-vb87c * Using SA for auth in tpu-tests * Updated SA with container tag for auth in tpu-tests * Added docker socket mount test * Updated tpu-tests to just pull and test the image from artifact registry after attaching service accounts as IAM policies * Added pytest command to the workflow * added grain installation command * Pruned unwanted files * included grain in requirements.txt * Updated tpu-tests.yml to use python image and explicitly install specific backend deps * Renamed tpu-tests to tpu-tests-jax and logging TPU device kind * Added a step to check gcloud installation * Running pytest on generic tpu workflow * Made changes as per suggestions in PR * Fixed error in action file * Added a job in tpu workflow to persist failed tests list * using requirements-jax-tpu.txt * Reverted the tpu-tests-jax.yml * Removed a command line option in tpu workflow file * Removed uninstall step from tpu workflow job * reverted tensorflow version in requirements file * Updated the tpu workflow to skip failing tests * Clean up TPU tests workflow by removing comment Removed commented out pip uninstall command from workflow. * Changed the failed tests file path and updated the same in conftest.py * Updated the failed test file path * Updated the workflow and job names for TPU * Added TPU job in actions.yml * Added TPU config print line * Added condition to TPU test job so that it gets triggered only after the PR review is approved * Added TPU specific tests in seperate workflow with PR approval condition * Removed pull_request condition for workflow execution and renamed the workflow --------- Co-authored-by: hertschuh <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
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.
Arbitrary functions and classes are not allowed.
OperationextendKerasSaveable, this required moving imports to avoid circular importsLayerno longer need to extendKerasSaveabledirectlyCrossandFeatureextendKerasSaveableenable_unsafe_deserialization