v2.0.3 #460
Merged
v2.0.3 #460
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
Raise awareness regarding missing features and incompatibility between the versions. Similar changes can be made for the docs.
…ent code Torch is very slow, so I had to increase the timeout accordingly.
* summary networks: add tests for using functional API * fix build functions for use with functional API
* fix docs of coupling flow * add additional tests
In addition, this PR limits the slow test to Windows and Python 3.10. The choices are somewhat arbitrary, my thought was to test the setup not covered as much through use by the devs.
…move update-workflows branch from workflow style tests, add __init__ and conftest to test_point_approximators (#443)
* implement compile_from_config and get_compile_config * add optimizer build to compile_from_config
* remove the is_symbolic_tensor check because this would otherwise skip the whole function for compiled contexts * skip pyabc test * fix sinkhorn and log_sinkhorn message formatting for jax by making the warning message worse
- use torch as default backend - reduce range of N so users of jax won't be stuck with a slow notebook - use BayesFlow built-in MLP instead of keras.Sequential solution - general code cleanup
amend to trigger workflow
* one place less to update * fix one link to a notebook
pip install -U pip setuptools wheel leads to an error: https://github.com/bayesflow-org/bayesflow/actions/runs/14692655483/job/41230057180?pr=449
Add advice regarding moving from v1 to v2 to README.
* minimal working case (.scale) * concatenate * keep, drop, rename * scale, log, sqrt * standardize * constraint transforms * continuous approximator returns log_prob with volume correction * loop for inverse jacobian * inverse for elementwise * inverse for Transforms * raise error with numpy transform (for now) * do not fail if no transform is used * take care of log1p as well * fix filter transforms, boundary condition * add tests for adapter jacobians * document jacobian arg * jacobian -> log_det_jac * add test for inverse concatenation * fix standardize * correct nesting in map_transform
…eveloper docs (#450) * document and expose serialization module * add functools.wraps call to allow_kwargs decorator, as before it was breaking the autodoc functionality * restructure and update developer docs * move content to separate pages * update section on serialization * ci: update pip via python -m pip pip install -U pip setuptools wheel leads to an error: https://github.com/bayesflow-org/bayesflow/actions/runs/14692655483/job/41230057180?pr=449 * serializable: increase depth in sys._getframe The functools.wrap decorator adds a frame object to the call stack * deprecate (de)serialize_value_or_type - add deprecation warning, remove functionality - replace all occurences with the corresponding new functions
- add `bootstrap_comparison` and `summary_space_comparison` to enable comparisons of two domains in the data space or the summary space via bootstrapping - add `.summaries()` function for easy access to summaries to `ContinuousApproximator` and `ModelComparisonApproximator` - add tests for the added functionality --------- Co-authored-by: Valentin Pratz <git@valentinpratz.de>
* v2.0.2 (#447) * [no ci] notebook tests: increase timeout, fix platform/backend dependent code Torch is very slow, so I had to increase the timeout accordingly. * Enable use of summary networks with functional API again (#434) * summary networks: add tests for using functional API * fix build functions for use with functional API * [no ci] docs: add GitHub and Discourse links, reorder navbar * [no ci] docs: acknowledge scikit-learn website * [no ci] docs: capitalize navigation headings * More tests (#437) * fix docs of coupling flow * add additional tests * Automatically run slow tests when main is involved. (#438) In addition, this PR limits the slow test to Windows and Python 3.10. The choices are somewhat arbitrary, my thought was to test the setup not covered as much through use by the devs. * Update dispatch * Update dispatching distributions * Improve workflow tests with multiple summary nets / approximators * Fix zombie find_distribution import * Add readme entry [no ci] * Update README: NumFOCUS affiliation, awesome-abi list (#445) * fix is_symbolic_tensor * remove multiple batch sizes, remove multiple python version tests, remove update-workflows branch from workflow style tests, add __init__ and conftest to test_point_approximators (#443) * implement compile_from_config and get_compile_config (#442) * implement compile_from_config and get_compile_config * add optimizer build to compile_from_config * Fix Optimal Transport for Compiled Contexts (#446) * remove the is_symbolic_tensor check because this would otherwise skip the whole function for compiled contexts * skip pyabc test * fix sinkhorn and log_sinkhorn message formatting for jax by making the warning message worse * update dispatch tests for more coverage * Update issue templates (#448) * Hotfix Version 2.0.1 (#431) * fix optimal transport config (#429) * run linter * [skip-ci] bump version to 2.0.1 * Update issue templates * Robustify kwargs passing inference networks, add class variables * fix convergence method to debug for non-log sinkhorn * Bump optimal transport default to False * use logging.info for backend selection instead of logging.debug * fix model comparison approximator * improve docs and type hints * improve One-Sample T-Test Notebook: - use torch as default backend - reduce range of N so users of jax won't be stuck with a slow notebook - use BayesFlow built-in MLP instead of keras.Sequential solution - general code cleanup * remove backend print * [skip ci] turn all single-quoted strings into double-quoted strings * turn all single-quoted strings into double-quoted strings amend to trigger workflow --------- Co-authored-by: Valentin Pratz <git@valentinpratz.de> Co-authored-by: Valentin Pratz <112951103+vpratz@users.noreply.github.com> Co-authored-by: stefanradev93 <stefan.radev93@gmail.com> Co-authored-by: Marvin Schmitt <35921281+marvinschmitt@users.noreply.github.com> * drafting feature * Initialize projectors for invariant and equivariant DeepSet layers * implement requested changes and improve activation --------- Co-authored-by: Lars <lars@kuehmichel.de> Co-authored-by: Valentin Pratz <git@valentinpratz.de> Co-authored-by: Valentin Pratz <112951103+vpratz@users.noreply.github.com> Co-authored-by: stefanradev93 <stefan.radev93@gmail.com> Co-authored-by: Marvin Schmitt <35921281+marvinschmitt@users.noreply.github.com>
* serialization: apply new scheme for `package` (breaking) - introduces new policy for consistent naming for serilization (see #451 for a discussion): standard is the path of a module a class resides in, trucated at depth to. So for all class in bayesflow.networks, we set package="bayesflow.networks", even if the live in the bayesflow.networks.mlp submodule. - The `serializable` decorator checks this and errors if this is not followed. The check can be disabled for certain cases (e.g., classes in the experimental module, that might eventually live somewhere else). - After this commit, previously saved models will not be loadable. As we introduced a bug regarding this anyway (#451), we will accept this and should inform users about it. - usage of direct calls to `keras.saving.register_keras_serializable` were replaced with our custom decorator. * update serilization policy in dev docs * README: add not regarding breaking changes until 2.1 release * standardize use of serializable decorator * [no ci] change (de)serialize to new pipeline in transform * serialization check: exempt classes not in bayesflow module This should ensure that users that try to use our decorator with external classes do not encounter the error. Possible edge case: they also name their module "bayesflow". --------- Co-authored-by: LarsKue <lars@kuehmichel.de>
# Conflicts: # bayesflow/links/ordered.py # bayesflow/metrics/maximum_mean_discrepancy.py # bayesflow/metrics/root_mean_squard_error.py # tests/test_workflows/conftest.py
|
I think it would be good to add that the new release is not compatible serialization-wise with the |
I would add it to the release note. |
|
I added a note to the PR comment too, nonetheless. |
vpratz
approved these changes
May 5, 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.
Important Note:
Due to changes in the serialization pipeline, this release is not backward-compatible with v2.0.2. You will likely not be able to load models created in v2.0.2 or lower.
This pull request introduces significant updates to the BayesFlow library, focusing on API enhancements, backward compatibility considerations, and improved functionality for transforms. The changes include updates to the documentation to guide users transitioning to version 2.0+, adjustments to the initialization process, and the addition of new features to various transform classes, such as support for log-determinant Jacobian calculations.
Documentation Updates
README.mdto inform users about breaking changes introduced in BayesFlow 2.0+, missing features (e.g., hierarchical models, MCMC), and guidance for migrating from version 1.x. [1] [2]Core Enhancements
__version__attribute initialization inbayesflow/__init__.pyby dynamically assigning it usingimportlib.metadata.version. Removed exception handling for unsupported packages.Transform Enhancements
log_det_jac(log-determinant of the Jacobian) calculations to several transform classes, includingConstrain,Concatenate, andDrop. This enables better handling of transformations in probabilistic models. [1] [2] [3]forwardandinversemethods inAdapterto optionally return the log-determinant Jacobian alongside transformed data. [1] [2]Serialization Improvements
@serializabledecorator across multiple transform classes (e.g.,AsSet,Broadcast,Constrain) to include a namespace argument ("bayesflow.adapters"), improving serialization consistency. [1] [2] [3]These changes collectively improve the usability, extensibility, and robustness of the BayesFlow library, particularly for users transitioning to version 2.0+ and those working with advanced transformations.