Skip to content

The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface. The fastest local inference engine in the world.

Commits on master

bb4416d
Missing imports.
comfyanonymous committed 2 years ago
c0b0da2
Move calculate function to comfy.lora
comfyanonymous committed 2 years ago
c26ca27
Code cleanups.
comfyanonymous committed 2 years ago
7c6bb84
Fix issue with models staying loaded in memory.
comfyanonymous committed 2 years ago
c54d3ed
Try to fix some lora issues.
comfyanonymous committed 2 years ago
c7ee4b3
Fix issue.
comfyanonymous committed 2 years ago
8f60d09
5f84ea6
fp16 is actually faster than fp32 on a GTX 1080.
comfyanonymous committed 2 years ago
843a7ff
Fix slow performance on 10 series Nvidia GPUs.
comfyanonymous committed 2 years ago
a60620d
Try a different type of flux fp16 fix.
comfyanonymous committed 2 years ago
015f73d
Make --fast work on pytorch nightly.
comfyanonymous committed 2 years ago
904bf58
Replace use of .view with .reshape (#4522)
Svein Ove Aas committed 2 years ago
5f50263
add a get models list api route (#4519)
Alex "mcmonkey" Goodwin committed 2 years ago
5e806f5
Add GET /internal/files. (#4295)
Robin Huang committed 2 years ago
f07e5bb
03ec517
Update frontend to 1.2.30 (#4513)
Chenlei Hu committed 2 years ago
bb50e69
Speed up fp8 matrix mult by using better code.
comfyanonymous committed 2 years ago
510f343
Simpletrainer lycoris format.
comfyanonymous committed 2 years ago
ea63b1c
9953f22
d1a6bd6
Make cast_to a nop if weight is already good.
comfyanonymous committed 2 years ago
538cb06
Fix potential issue with multi devices.
comfyanonymous committed 2 years ago
1b3eee6
Update README.md (Add shield badges) (#4490)
Chenlei Hu committed 2 years ago
5a69f84
New load_text_encoder_state_dicts function.
comfyanonymous committed 2 years ago
9eee470