Releases: juglab/FourierImageTransformer
Releases · juglab/FourierImageTransformer
v0.2.0
Refactored code-base and updated jupyter examples.
v0.1.24_zero
This is the release with the final code version used for the manuscript.
v0.1.20
- Refactor SRes and add fast SRes prediction.
- Change loss to use sum instead of product.
- Add crop lodopab.
v0.1.19
- Fix major bug in positional encoding
- Add polar coordinate positional embedding
v0.1.18
Merge pull request #14 from juglab/merger
v0.1.18
v0.1.17
Divide bin-factor by 2 instead of reducing it by 1.
v0.1.16
Merge pull request #12 from juglab/merger
v0.1.16
v0.1.10
Compute amplitude-loss directly on network output i.e. do not denormalize it.
v0.1.8
Make first conv in conv-block have kernel size 3.