James Betker 812c684f7d Update pixgan swap algorithm ... - Swap multiple blocks in the image instead of just one. The discriminator was clearly learning that most blocks have one region that needs to be fixed. - Relax block size constraints. This was in place to gaurantee that the discriminator signal was clean. Instead, just downsample the "loss image" with bilinear interpolation. The result is noisier, but this is actually probably healthy for the discriminator.		2020-07-10 15:56:14 -06:00
..
archs	Move ExpansionBlock to arch_util	2020-07-10 15:53:41 -06:00
__init__.py	Implement downsample GAN	2020-04-24 00:00:46 -06:00
base_model.py	Fix inverse temperature curve logic and add upsample factor	2020-06-19 09:18:30 -06:00
loss.py	Fix up pixgan loss and pixdisc	2020-07-08 21:27:48 -06:00
lr_scheduler.py	Enable forced learning rates	2020-06-07 16:56:05 -06:00
networks.py	SRG2 revival	2020-07-09 17:34:51 -06:00
SR_model.py	Allow validating in batches, remove val size limit	2020-06-02 08:41:22 -06:00
SRGAN_model.py	Update pixgan swap algorithm	2020-07-10 15:56:14 -06:00