T

Jukka Seppänen 265f46b54b Merge pull request #4 from gluttony-10/main

Update pyramid_dit_for_video_gen_pipeline.py

2024-10-11 09:09:23 +03:00

diffusion_schedulers

fp8_fast mode, cleanup

2024-10-11 00:47:08 +03:00

examples

img2vid example update

2024-10-11 02:22:08 +03:00

pyramid_dit

Update pyramid_dit_for_video_gen_pipeline.py

2024-10-11 10:25:45 +08:00

trainer_misc

fp8_fast mode, cleanup

2024-10-11 00:47:08 +03:00

video_vae

fp8_fast mode, cleanup

2024-10-11 00:47:08 +03:00

__init__.py

init

2024-10-10 13:16:35 +03:00

.gitattributes

Initial commit

2024-10-10 13:12:28 +03:00

.gitignore

init

2024-10-10 13:16:35 +03:00

fp8_optimization.py

fp8_fast mode, cleanup

2024-10-11 00:47:08 +03:00

LICENSE

init

2024-10-10 13:16:35 +03:00

nodes.py

Update nodes.py

2024-10-11 02:12:36 +03:00

README.md

don't download both models

2024-10-11 01:56:46 +03:00

requirements.txt

update flash_attn and reqs

2024-10-10 22:10:45 +03:00

utils.py

init

2024-10-10 13:16:35 +03:00

README.md

ComfyUI wrapper nodes for Pyramid-Flow

NOT FULLY FUNCTIONAL YET

todo

refactor to use comfy text encoding instead
optimize memory use

Besides text encoder, this should already run at 10-12GB VRAM

Model loading has not been optimized at all yet, currently needs everything from here:

https://huggingface.co/rain1011/pyramid-flow-sd3/tree/main

to:

ComfyUI/models/pyramidflow/pyramid-flow-sd3

Original repo: https://github.com/jy0205/Pyramid-Flow