LoRA Fine-Tuning Script for Single-GPU (24 GB) Setups

[joeyu930]

Hi,

I’d like to fine-tune StableAnimator on a single RTX 3090 Ti (24 GB VRAM). The README notes that full fine-tuning currently requires around 40 GB of VRAM, so I’m writing to ask:

1. Is there already a LoRA / low-rank adaptation fine-tuning script or branch available?

2. If not yet public, could you share any work-in-progress implementation or an estimated release plan?

Thank you for the excellent work—looking forward to your reply!