chengzeyi/stable-fast
View on GitHub ↗ ★ 1,306 Python
https://wavespeed.ai/ Best inference performance optimization framework for HuggingFace Diffusers on NVIDIA GPUs.
6 mentions on Hacker News by 2 people: 2 stories, 4 comments · first 2023-10-17 · latest 2023-11-30
20082026
-
▲ 2
Stable-fast for SD inference: Faster than AITemplate, On par with TensorRTstory by chengzeyi · 2023-11-30 · 0 comments
-
SDXL and ControlNet are already optimized, if thats what you mean: https://github.com/chengzeyi/stable-fast (Note the links to various SD compilers). But the whole field is moving so fast that people aren't even adopting the compilers and…
comment by brucethemoose2 · 2023-11-29 · in SDXL Turbo: A Real-Time Text-to-Image Generation Model -
…still check out stable-fast, which is claiming ~16ms/iter on a 4090, twice that of torch.compile: https://github.com/chengzeyi/stable-fast#rtx-4090-512x512-ba...
comment by brucethemoose2 · 2023-11-13 · in Show HN: Realtime LCM tool -
Now combine this with an optimized SD implementation, like: https://github.com/chengzeyi/stable-fast Or AITemplate, and you are at 15FPS on a larger consumer GPU. 10 with a controlnet you can use for some motion consistency.
comment by brucethemoose2 · 2023-11-13 · in Generate images fast with SD 1.5 while typing on Gradio -
Since I am sending you down the rabbit hole anyway, you should check out stable-fast: https://github.com/chengzeyi/stable-fast It's the most promising "fast" and flexible stable diffusion implementation akin to this paper or vLLM that I…
comment by brucethemoose2 · 2023-11-09 · in S-LoRA: Serving Concurrent LoRA Adapters -
▲ 2
Stable Fast: Lightweight Inference Optimization Library for Stable Diffusionstory by chengzeyi · 2023-10-17 · 1 comment