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

Browse Source 
Signed-off-by: Vladimir Mandic <mandic00@live.com>
This commit is contained in:
Vladimir Mandic
2024-10-22 15:54:21 -04:00
parent f191134aa6
commit b28b8403b7
1 changed files with 13 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files
CHANGELOG.md
+13 -7
View File
@@ -17,7 +17,7 @@
#### Newly supported
- New fine-tuned [CLiP-ViT-L]((https://huggingface.co/zer0int/CLIP-GmP-ViT-L-14)) 1st stage **text-encoders** used by SD15, SDXL, Flux.1, etc. brings additional details to your images
- New fine-tuned [CLiP-ViT-L]((https://huggingface.co/zer0int/CLIP-GmP-ViT-L-14)) 1st stage **text-encoders** used by most models (SD15/SDXL/SD3/Flux/etc.) brings additional details to your images
- New models:
[Stable Diffusion 3.5 Large](https://huggingface.co/stabilityai/stable-diffusion-3.5-large)
[OmniGen](https://arxiv.org/pdf/2409.11340)
@@ -34,8 +34,8 @@
- Auto-detection of best available **device/dtype** settings for your platform and GPU reduces neeed for manual configuration
- Full rewrite of **sampler options**, not far more streamlined with tons of new options to tweak scheduler behavior
- Improved **LoRA** detection and handling for all supported models
- Tons of work on **dynamic quantization** that can be applied on-the-fly during model load to any model type
Supported quantization engines include `TorchAO`, `Optimum.quanto`, `NNCF` compression, and more...
- Tons of work on **dynamic quantization** that can be applied *on-the-fly* during model load to any model type (*you do not need to use pre-quantized models*)
Supported quantization engines include `BitsAndBytes`, `TorchAO`, `Optimum.quanto`, `NNCF` compression, and more...
Oh, and we've compiled a full table with list of top-30 (*how many have you tried?*) popular text-to-image generative models,
their respective parameters and architecture overview: [Models Overview](https://github.com/vladmandic/automatic/wiki/Models)
@@ -249,10 +249,16 @@ And there are also other goodies like multiple *XYZ grid* improvements, addition
- setting `lora_load_gpu` to load LoRA directly to GPU
*default*: true unless lovwram
- **torchao**
- reimplement torchao quantization
- configure in settings -> compute settings -> quantization
- can be applied to any model on-the-fly during load
- **quantization**
- new top level settings group as we have quite a few quantization options now!
configure in *settings -> quantization*
- in addition to existing `optimum.quanto` and `nncf`, we now have `bitsandbytes` and `torchao`
- **bitsandbytes**: fp8, fp4, nf4
- quantization can be applied on-the-fly during model load
- currently supports `transformers` and `t5` in **sd3** and **flux**
- **torchao**: int8, int4, fp8, fp4, fpx
- configure in settings -> quantization
- can be applied to any model on-the-fly during load
- **huggingface**:
- force logout/login on token change