Aidan Dempster c85e0d5c35 Update decoder dataloader (#105) ...

* Updated the decoder dataloader
Removed unnecessary logging for required packages
Transferred to using index width instead of shard width
Added the ability to select extra keys to return from the webdataset

* Added README for decoder loader

2022-05-20 16:38:55 -07:00

..

data

bring in the simple tokenizer released by openai, but also plan on leaving room for custom tokenizer with yttm

2022-04-12 09:23:17 -07:00

dataloaders

Update decoder dataloader (#105)

2022-05-20 16:38:55 -07:00

__init__.py

unet_number on decoder trainer only needs to be passed in if there is greater than 1 unet, so that unconditional training of a single ddpm is seamless (experiment in progress locally)

2022-05-16 09:17:17 -07:00

cli.py

add missing import (#56)

2022-05-04 07:42:20 -07:00

dalle2_pytorch.py

quick fix for @marunine

2022-05-18 20:22:52 -07:00

optimizer.py

default decoder learning rate to what was in the paper

2022-05-16 13:33:54 -07:00

tokenizer.py

bring in the simple tokenizer released by openai, but also plan on leaving room for custom tokenizer with yttm

2022-04-12 09:23:17 -07:00

trackers.py

experiment tracker agnostic

2022-05-15 09:56:40 -07:00

trainer.py

update sample, and set default gradient clipping value for decoder training

2022-05-16 17:38:30 -07:00

vqgan_vae_trainer.py

unet_number on decoder trainer only needs to be passed in if there is greater than 1 unet, so that unconditional training of a single ddpm is seamless (experiment in progress locally)

2022-05-16 09:17:17 -07:00

vqgan_vae.py

remove convnext blocks, they are illsuited for generative work, validated by early experimental results at https://github.com/lucidrains/video-diffusion-pytorch

2022-05-05 07:07:21 -07:00