Skip to content

Releases: HyperGAN/HyperGAN

HyperGAN 1.0 RC 2

09 Aug 13:41
722c082
Compare
Choose a tag to compare

Changelog:

  • Fixes windows missing hypergan.cmd(thanks @IIIBlueberry)
  • Fixed documentation links(thanks @AI-Spawn)
  • Better default configuration
  • General cleanup of unused functionality
  • Require hypergan version in configuration
  • Update colorizer configuration
  • New colorizer/super-resolution configuration
  • Classification example configuration
  • Fix for --crop when width and height were different
  • Fix for cat layer output shape
  • pip package continues to be updated

HyperGAN 1.0 RC 1

02 Aug 00:42
Compare
Choose a tag to compare

HyperGAN 1.0 RC 1

HyperGAN 1.0 is the result of roughly one year of directed effort to stabilize and speed up training.

When we started the project we decided that 1.0 was going to be released when we have a stable convergent default. Now that's here we're excited to bring you HyperGAN 1.0!

With this release we are also announcing the github sponsorship program. Help us develop HyperGAN into the GAN framework we know it can be!

https://github.com/sponsors/hypergan

If you have tried HyperGAN before, please give it another shot.

If you are new to HyperGAN, welcome, let us know more about your use case in the discord.

Changes

  • Hypergan 1.0 now uses Pytorch instead of Tensorflow 1.0!
  • New gitbook documentation! https://hypergan.gitbook.io/hypergan/
  • New formalized HyperGAN layer definition language for defining your own network layouts
  • New stabilization technique independently discovered and not available in other frameworks. See the default configuration for more.
  • UI improvements
  • Faster training
  • WAY faster startup time
  • More robust input handling - 0 byte files are not a problem!
  • --crop now does a resize THEN a crop to capture as much of the target image as possible.
  • Fix samplers and enable more of them
  • Switch samplers while training through menu
  • New --width for specifying number sampler columns
  • Updated configurations
  • New working default configuration
  • New AlignedGAN for aligning one distribution to another
  • New AlignedInterpolatedGAN for discovering novel distributions
  • Added TOML support ( #240 )
  • Lots of new layers, including exciting layers such as multi_head_attention. See the (auto-generated) documentation for more.
  • Unique layers not available in other frameworks, such as segment_softmax which increases generator capacity significantly
  • Much much more...

We've tried to keep the good things about hypergan and remove the gnarly parts. We're still documenting and improving things but it's working and ready to be tested out.

To be continued...

We still have a lot of things planned for this release. Consider this an open invitation to start seeing experimenting with 1.0 and let us know what you think in the discord!

HyperGAN 0.10.1

28 Jul 04:23
0c68bd7
Compare
Choose a tag to compare

bug fixes

HyperGAN 0.10.0

27 Jul 07:05
Compare
Choose a tag to compare

HyperGAN 0.10.0 is here

We're proud to announce that HyperGAN 0.10.0 has been officially released!

HyperGAN 0.10 introduces a lot. The end result is better generators with more variety on less hardware.

New features

Here's some of the features in hypergan:

  • HIP(HyperGAN Improvement Proposal) framework allows community participation in the roadmap
  • A new default configuration with lots of diversity that scales to any size
  • A 'configurable component' which allows you to define your model architecture in JSON files
  • New viewer
  • New android app with source code(release upcoming)
  • Support for 1(b&w), 3(rgb) or 4(rgba) channels
  • Loads more. New trainers, optimizers, components, new everything.

Additionally we've optimized. hypergan build can take a model that trains at 6gb and reduce it to < 10 MB at inference time. Suitable for embedding in mobile devices or web pages.

Upgrading

Make sure to uninstall previous versions of hypergan then pip install hypergan. On windows? Check the discord for instructions.

Join us

Join the discord where you can find exclusive content.

Thinking about helping out? With the release of 0.10.0 and HIPs now is the perfect time. Create a HIP. Create a model. Help with an issue. Share a dataset. Show your project off. There's lots of fun to be had and cool things to build.

More to come

We have several new features we're excited to share with you in 0.10. Stay tuned!

Social media

Follow us on twitter. Like and subscribe to the youtube channel where we will be releasing new videos. Star the github. And of course, join our discord if you haven't already. It really does help a lot.

Thank you

We'd like to thank everyone in the hypergan community for testing, supporting, suggesting, building, and generally being great during this development cycle.

HyperGAN 0.9.5

24 Jul 11:28
Compare
Choose a tag to compare
  • Fix to cross platform pygame viewer(thanks @sbuckleybonanno)
  • hypergan build builds .pbgraph file
  • documentation updates

0.9.4 - cross platform fixes

11 Jul 07:11
Compare
Choose a tag to compare

0.9.4

  • introduces support for python versions less than 3.5 (thanks @andrewyang96 )
  • Adds a cross platform viewer using pygame(tested on osx and linux, should work on windows)

0.9 released

11 Jul 07:13
Compare
Choose a tag to compare

0.9

This release includes huge refactorings and tons of new features. There are too many changes to list here, see #66 for full details