Releases: AllenInstitute/deepinterpolation
Releases · AllenInstitute/deepinterpolation
0.2.0
What's Changed
- CLI was removed. Some data passing layers caused multi-threading conflicts/garbage collection which prevented thread from closing when finished.
- Package import was greatly simplified to facilitate use.
- JSON intermediate object was removed. This facilitates direct access to class properties and make more conventional imports.
- Older examples and code related to experimentation for the publication was removed.
- Tests and examples were updated to match new syntax.
- Previous use of this package through the CLI need to update to the new python syntax that directly use the generator classes. This is explained in the example folder.
0.1.5
What's Changed
- Pass same multiprocessing settings to evaluate as to fit by @aamster in #87
- move to predict_on_batch by @jeromelecoq in #78
0.1.4
- This release contains the first version of the CLI along with documented examples.
- We updated requirements to fix a tensorflow critical issues and bring tensorflow to 2.4.4.
- We removed unnecessary version lock of required packages.
- Fixes a bug with CollectorGenerator that would loop over the data instead of continually providing new batches
- Refactored SequentialGenerator to have consistent parameters and behavior around start and end_frames for all temporal generator (2p, ephys, movies...).
- We published the package to PyPi for the first time.
Peer-reviewed publication
This release was finalized at the time of publication of the DeepInterpolation manuscript in a peer-reviewed journal.
Please disregard tag number 1.0. this was saved by mistake and is incompatible with the currently release number on PyPi. This version should be 0.1.x