Sign In

The pruned stable diffusion XL (SDXL 0.9) no-ema (for lower RAM and lower VRAM)

Updated: Oct 5, 2024

base model

Download

1 variant available

fp16 SafeTensor

sdxl09_pruned_base.safetensors

Half precision, best balance (pruned) • 6.46 GB

Verified:

Type

Checkpoint Merge

Stats

1,618

Reviews

Published

Jul 29, 2023

Base Model

SDXL 0.9

Hash

AutoV2
A7D2D9A924
default creator card background decoration
Followers - 518

518

Likes - 8931

8.9K

00002-4029279845.png

Just someone asked me to prune it...

The SDXL 0.9 version better for low RAM and VRAM, with some quality loss.

This is the base model, the refiner is not need to prune.

The same setting from the model without prune: