Update dev/main #276
Open
Update dev/main #276
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
This reverts commit 7559b5c.
Let's specify resolution also directly like `(8, 8, 8)`, in addition to
`Coordinate(8, 8, 8)`?
```python
datasplit_config = DataSplitGenerator.generate_from_csv(
'test.csv',
input_resolution=(8, 8, 8), # This works.
output_resolution=Coordinate(4, 4, 4), # And this works.
)
```
There appear to be some python formatting errors in 330365b. This pull request uses the [psf/black](https://github.com/psf/black) formatter to fix these issues.
previously file stat writing would overwrite existing stats; this fix prevents that by appending the new stats
This PR adds an optional `augmentation_probability: float = 1.` argument to `ElasticAugment`, `IntensityAugment`, `SimpleAugment`.
fixing batch dim bugs (batch norm requires batch dimension even in predict mode) Seems to also fix the strange loss spike. I think it was due to setting model into eval mode and then not resetting to training at the end
Exceptions DVID and Resampled arrays
Upgrade to funlib.persistence `0.5`. This update makes a one big improvement: Custom `Array` class no longer needed. We used this mostly just to apply preprocessing lazily to large arrays. New `funlib` `Array` class uses `dask` internally which comes with much better support for lazy array operations than we built for ourselves. The `ZarrArray` and `NumpyArray` class which were used extensively throughout `DaCapo` have now been replaced with simple `funlib.persistence.Array`s. A minor incompatibility: `funlib.persistence.Array` has a convention (for now) that all axes have names, but non-spatial axes have a "^" in their name. This will be fixed in the near future. For now, DaCapo convention needed to change a little bit to adapt to this. We now have to use "c^" and "b^" for channel and batch dimensions instead of just "c" and "b". TODOs: This pull request is not quire ready to merge. I pass the tests run with `pytest`, and the `minimal_tutorial` notebook executes. But there is a lot of code that is not tested. Specifically many of the `ArrayConfig` subclasses are not yet tested and some are missing implementations. Here are the Preprocessing array configs, whether or not their implementation is complete, and their code coverage: - [X] BinarizeArrayConfig 96% - [X] ConcatArrayConfig 60% - [X] ConstantArrayConfig 57% - [X] CropArrayConfig 69% - [X] DummyArrayConfig 91% - [ ] DVIDArrayConfig 90% (misleading, only skeleton implementation so not much to test) - [X] IntensitiesArrayConfig 75% - [X] LogicalOrArrayConfig 60% - [x] MergeInstancesArrayConfig 100% (misleading, no implementation so nothing to test) - [x] MissingAnnotationsMaskConfig 100% (misleading) - [x] OnesArrayConfig 100% (misleading) - [ ] ResampledArrayConfig 100% (misleading) - [x] SumArrayConfig 100% (misleading) - [x] TiffArrayConfig 0% - [X] ZarrArrayConfig 70% Best practice would be to add tests before merging, but I want to put this here so others can test it
There appear to be some python formatting errors in aeb77a6. This pull request uses the [psf/black](https://github.com/psf/black) formatter to fix these issues.
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.
No description provided.