tingxueronghua 9ed87aa688 support fp8 model loading and cpu offload strategy. Thus the workflow could run in 24GB GPU.

2025-04-30 14:54:42 +08:00

..

__init__.py

init repo

2025-04-26 23:20:54 +08:00

attention.py

init repo

2025-04-26 23:20:54 +08:00

autoencoder.py

init repo

2025-04-26 23:20:54 +08:00

conditioner.py

init repo

2025-04-26 23:20:54 +08:00

connector_edit.py

support fp8 model loading and cpu offload strategy. Thus the workflow could run in 24GB GPU.

2025-04-30 14:54:42 +08:00

layers.py

support fp8 model loading and cpu offload strategy. Thus the workflow could run in 24GB GPU.

2025-04-30 14:54:42 +08:00

model_edit.py

init repo

2025-04-26 23:20:54 +08:00