upstream-archive/stable-diffusion.cpp
1
0
Fork 0
mirror of https://github.com/leejet/stable-diffusion.cpp.git synced 2026-07-23 11:20:53 -05:00
Code Issues Packages Projects Releases Wiki Activity

feat: add spectrum caching method (#1322)

Browse Source
This commit is contained in:
rmatif
2026-03-09 17:35:32 +01:00
committed by GitHub
parent c8fb3d2458
commit dea4980f4e
6 changed files with 325 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

23
docs/caching.md
View File

@@ -11,6 +11,7 @@ Caching methods accelerate diffusion inference by reusing intermediate computati
| `dbcache` | DiT models | Block-level L1 residual threshold |
| `taylorseer` | DiT models | Taylor series approximation |
| `cache-dit` | DiT models | Combined DBCache + TaylorSeer |
| `spectrum` | UNET models | Chebyshev + Taylor output forecasting |
### UCache (UNET Models)
@@ -118,6 +119,28 @@ Mask values: `1` = compute, `0` = can cache.
--scm-policy dynamic
```
### Spectrum (UNET Models)
Spectrum uses Chebyshev polynomial fitting blended with Taylor extrapolation to predict denoised outputs, skipping entire UNet forward passes. Based on the paper [Spectrum: Adaptive Spectral Feature Forecasting for Efficient Diffusion Sampling](https://github.com/tingyu215/Spectrum).
```bash
sd-cli -m model.safetensors -p "a cat" --cache-mode spectrum
```
#### Parameters
| Parameter | Description | Default |
|-----------|-------------|---------|
| `w` | Chebyshev vs Taylor blend weight (0=Taylor, 1=Chebyshev) | 0.40 |
| `m` | Chebyshev polynomial degree | 3 |
| `lam` | Ridge regression regularization | 1.0 |
| `window` | Initial window size (compute every N steps) | 2 |
| `flex` | Window growth per computed step after warmup | 0.50 |
| `warmup` | Steps to always compute before caching starts | 4 |
| `stop` | Stop caching at this fraction of total steps | 0.9 |
```
### Performance Tips
- Start with default thresholds and adjust based on output quality