{{ message }}
Repository navigation
Analysis of Classifier-Free Guidance Weight Schedulers #7756
Description
Model/Pipeline/Scheduler description
The paper's authors performs an analysis and proposes 1 line change in order to make Classifier-Free Guidance looks better
I personally run some test to confirm
SD1.5 DDIM scheduler, 50 steps, "a photograph of an astronaut riding a horse", seed: 1024
guidance scale: 7.5
static
linear (proposed)
guidance scale: 14.0
static
linear (proposed)
Open source status
- The model implementation is available.
- The model weights are available (Only relevant if addition is not a scheduler).
Activity
rootonchair commentedon Apr 23, 2024
Update SDXL result
guidance scale: 14.0
static
linear
DN6 commentedon Apr 24, 2024
cc: @yiyixuxu and @asomoza for visibility.
bghira commentedon Apr 26, 2024
what a neat way to make use of the knowledge already in the model!
yiyixuxu commentedon Apr 26, 2024
cc @asomoza can we make a callback for this?
asomoza commentedon Apr 26, 2024
yes but there's a lot of techniques about manipulating the
CFG, most of them without papers, I added the cutout one because I know it's really popular, makes the generations faster, and as a kind of example on how to manipulate theCFGMaybe we should let the community add these ones later on? Other more popular ones are automatic cfg and Dynamic Thresholding
YunhoKim21 commentedon May 16, 2024
Hi, interested on your work. Is there an explanation of WHY this phenomena happens?
bghira commentedon May 16, 2024
i'm not an expert on this, but on a cursory glance it seems to be basing the strength of guidance by the position in the timestep schedule. this also likely works because there's two types of attention being used by the model, with earlier timesteps being cross-attn (heavily relying on text conditional input) and later timesteps being self-attn (practically ignoring the prompt)
github-actions commentedon Sep 14, 2024
This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.
Please note that issues that do not follow the contributing guidelines are likely to be ignored.
sayakpaul commentedon Nov 30, 2024
@yiyixuxu @asomoza @a-r-r-o-w this could be a good candidate for modular diffusers no?
bghira commentedon Nov 30, 2024
modular diffusers?
asomoza commentedon Nov 30, 2024
yes, actually all kind of variations the the CFG and also all the techniques that use the callbacks are the best initial candidates for the modular diffusers.
github-actions commentedon Dec 25, 2024
This issue has been automatically marked as stale because it has not had recent activity. If you think this still needs to be addressed please comment on this thread.
Please note that issues that do not follow the contributing guidelines are likely to be ignored.