mengqin

SageAttention. [ICLR2025, ICML2025, NeurIPS2025 Spotlight] Quantized Attention achieves speedup of 2-5x compared to FlashAttention, without losing end-to-end metrics across language, image, and video models.

ComfyUI-TwinFlow. A ComfyUI custom node implementation of TwinFlow: Realizing One-step Generation on Large Models with Self-adversarial Flows.

ComfyUI-UnetBnbModelLoader. A general comfyui model loading plugin that supports loading unet models quantized in bnb-4bit (nf4 and fp4) format

MSMF. M-Simple Makefile Framework

MIUI-12.5-GMS-permissions-files. MIUI 12.5 CN image lack some important permissions files to let GMS work correct, here's it's, extract them from Global image.