Hypergan

* Switched to pytorch
* Almost a complete rewrite
* Faster, cleaner, more stable
* Model definition language formalized
* New custom train hooks not available in other projects
* More...

* Initial private release

 

* Configurable network architectures using a simple DSL
* Lots of new regularizers and losses
* Evolution based GangTrainer
* Curriculum trainer with progressive growing
* Optimistic loading
* New next-frame example
* More...

* API Documentation - [https://s3.amazonaws.com/hypergan-apidocs/0.9.0/index.html](https://s3.amazonaws.com/hypergan-apidocs/0.9.0/index.html)
* Prepackaged configurations
* New viewer front-end!
* Examples, including the ability to randomly search for good configurations

See more here [https://github.com/255BITS/HyperGAN/pull/66](https://github.com/255BITS/HyperGAN/pull/66)

* Tensorflow 1.0 support
* New configuration format and refactored api.
* New loss based on least squared GAN. See [least squares GAN implementation](changelog.mdLeast-Squares-GAN).
* API example `2d-test` - tests a trainer/encoder/loss combination against a known distribution.
* API example `2d-measure` - measure and report the above test by randomly combining options.
* Updated default configuration.
* More

* New loss `wgan`
* Initial Public API Release
* API example: `colorizer` - re-colorize an image!
* API example: `inpainter` - remove a section of an image and have your GAN repaint it
* API example: `super-resolution` - zoom in and enhance. We've caught the bad guy!
* 4 _new_ samplers. `--sampler` flag. Valid options are: `batch`,`progressive`,`static_batch`,`grid`.