Downloadable VRAM!
Qwen Edit Lightning loras with lower ranks that take up less space, when VRAM space is at a premium, use these!
In my opinion v1.1 is the best version, followed by 2.0 and then 1.0
Description
Rank 32
FAQ
Comments (8)
what's your workflow for down converting these? I'd like to try the downconvert on some of my trained QWEN loras.
https://huggingface.co/anyMODE/python-script-dump/blob/main/lora_redim.py I used this script the randomized_svd mode is very fast
Can you do this for edit 2509?
Are there lightning loras specifically for that?
alucardx651Oct 17, 2025
@WANdalf Hahaha you was fast, thank you ;)
Sorry for noon question, what is the difference in R32, R16 and R8 versions of 1.1? Quality, sharpness, speed?
They're sizes, you get the best quality from the highest rank usually