Public/automatic
1
0
Fork 0
mirror of https://github.com/vladmandic/automatic synced 2026-09-19 01:04:32 +02:00
Code Issues Packages Projects Releases Wiki Activity

update changelog and cleanup

Browse Source 
Signed-off-by: Vladimir Mandic <mandic00@live.com>
This commit is contained in:
Vladimir Mandic
2025-03-27 11:53:03 -04:00
parent 2c58d3b36c
commit 2a18890235
2 changed files with 23 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files
CHANGELOG.md
+20 -20
View File
@@ -62,16 +62,6 @@ Pretty big performance updates to a) Any model using DiT based architecture: new
download text encoders into folder set in settings -> system paths -> text encoders (default is *models/Text-encoder*)
load using *settings -> text encoder*
*tip*: add *sd_text_encoder* to your *settings -> user interface -> quicksettings* list to have it appear at the top of the ui
- **Wiki/Docs**
- updated [Models](https://github.com/vladmandic/sdnext/wiki/Models) info
- new [Video](https://github.com/vladmandic/sdnext/wiki/Video) guide
- new [Caption](https://github.com/vladmandic/sdnext/wiki/Caption) guide
- new [VAE](https://github.com/vladmandic/sdnext/wiki/VAE) guide
- updated [SD3](https://github.com/vladmandic/sdnext/wiki/SD3) guide
- updated [ZLUDA](https://github.com/vladmandic/sdnext/wiki/ZLUDA) guide
- updated [OpenVINO](https://github.com/vladmandic/sdnext/wiki/OpenVINO) guide
- updated [AMD-ROCm](https://github.com/vladmandic/sdnext/wiki/AMD-ROCm) guide
- upte [Intel-ARC](https://github.com/vladmandic/sdnext/wiki/Intel-ARC) guide
- **Acceleration**
- Support for most DiT-based models, for example: *FLUX.1, SD35, Hunyuan, Mochi, Latte, Allegro, Cog*
- Enable and configure in *Settings -> Pipeline modifiers*
@@ -94,16 +84,6 @@ Pretty big performance updates to a) Any model using DiT based architecture: new
- add image policy checks using `LlavaGuard` VLM to detect policy violations (and reasons)
against top-10 standard harmful content categories
- add banned words/expressions check against prompt variations
- **Other**
- **upscale**: new [asymmetric vae v2](https://huggingface.co/Heasterian/AsymmetricAutoencoderKLUpscaler_v2) upscaling method
- **upscale**: new experimental support for `libvips` upscaling
- **quantization**: add support for `optimum-quanto` on-the-fly quantization during load for all models
note: previous method for quanto is still valid and is noted in settings as post-load quantization
- add quantization support to **CogView-3Plus**
- update `diffusers` and other requirements
- rename vae, unet and text-encoder settings *None* to *Default* to avoid confusion
- **CLI**: add `cli/api-grid.py` which can generate grids using params-from-file for x/y axis
- LoRA enable memory cache by default
- **IPEX**
- add `--upgrade` to torch_command when using `--use-nightly`
- add xpu to profiler
@@ -118,6 +98,26 @@ Pretty big performance updates to a) Any model using DiT based architecture: new
- add `torch.compile` support under [triton for ZLUDA v3.9.2](https://github.com/vladmandic/sdnext/wiki/ZLUDA#how-to-enable-triton)
- add Flash Attention 2 support under [triton for ZLUDA v3.9.2](https://github.com/vladmandic/sdnext/wiki/ZLUDA#how-to-enable-triton)
- add Sage Attention support
- **Other**
- **upscale**: new [asymmetric vae v2](https://huggingface.co/Heasterian/AsymmetricAutoencoderKLUpscaler_v2) upscaling method
- **upscale**: new experimental support for `libvips` upscaling
- **quantization**: add support for `optimum-quanto` on-the-fly quantization during load for all models
note: previous method for quanto is still valid and is noted in settings as post-load quantization
- add quantization support to **CogView-3Plus**
- update `diffusers` and other requirements
- rename vae, unet and text-encoder settings *None* to *Default* to avoid confusion
- **CLI**: add `cli/api-grid.py` which can generate grids using params-from-file for x/y axis
- LoRA enable memory cache by default
- **Wiki/Docs**
- updated [Models](https://github.com/vladmandic/sdnext/wiki/Models) info
- new [Video](https://github.com/vladmandic/sdnext/wiki/Video) guide
- new [Caption](https://github.com/vladmandic/sdnext/wiki/Caption) guide
- new [VAE](https://github.com/vladmandic/sdnext/wiki/VAE) guide
- updated [SD3](https://github.com/vladmandic/sdnext/wiki/SD3) guide
- updated [ZLUDA](https://github.com/vladmandic/sdnext/wiki/ZLUDA) guide
- updated [OpenVINO](https://github.com/vladmandic/sdnext/wiki/OpenVINO) guide
- updated [AMD-ROCm](https://github.com/vladmandic/sdnext/wiki/AMD-ROCm) guide
- upte [Intel-ARC](https://github.com/vladmandic/sdnext/wiki/Intel-ARC) guide
- **Fixes**
- fix installer not starting when older version of `rich` is installed
- fix circular imports when debug flags are enabled
modules/transformer_cache.py
+3 -4
View File
@@ -11,10 +11,10 @@ def set_cache(faster_cache=None, pyramid_attention_broadcast=None):
return
faster_cache = faster_cache if faster_cache is not None else shared.opts.faster_cache_enabled
pyramid_attention_broadcast = pyramid_attention_broadcast if pyramid_attention_broadcast is not None else shared.opts.pab_enabled
if not faster_cache and not pyramid_attention_broadcast:
if (not faster_cache) and (not pyramid_attention_broadcast):
return
if not hasattr(shared.sd_model.transformer, 'enable_cache') or not hasattr(shared.sd_model.transformer, 'disable_cache'):
shared.log.debug(f'Transformer cache: cls={shared.sd_model.transformer.__class__.__name__} not supported')
if (not hasattr(shared.sd_model.transformer, 'enable_cache')) or (not hasattr(shared.sd_model.transformer, 'disable_cache')):
shared.log.debug(f'Transformer cache: cls={shared.sd_model.transformer.__class__.__name__} fc={faster_cache} pab={pyramid_attention_broadcast} not supported')
return
try:
if faster_cache: # https://github.com/huggingface/diffusers/pull/10163
@@ -32,7 +32,6 @@ def set_cache(faster_cache=None, pyramid_attention_broadcast=None):
shared.sd_model.transformer.disable_cache()
shared.sd_model.transformer.enable_cache(config)
shared.log.debug(f'Transformer cache: type={config.__class__.__name__}')
shared.log.critical(f'HERE: {vars(config)}')
debug(f'Transformer cache: {vars(config)}')
elif pyramid_attention_broadcast: # https://github.com/huggingface/diffusers/pull/9562
config = diffusers.PyramidAttentionBroadcastConfig(