Type | |
Stats | 6,005 |
Reviews | (155) |
Published | Jul 30, 2023 |
Base Model | |
Hash | AutoV2 1FA5725F4F |
This is merge model for:
1. 100% stable-diffusion-xl-base-1.0 and 100% stable-diffusion-xl-refine-1.0
https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0
https://huggingface.co/stabilityai/stable-diffusion-xl-refiner-1.0
2. sdxl-vae-fp16-fix
https://huggingface.co/madebyollin/sdxl-vae-fp16-fix
you can use this directly or finetune.
same license on stable-diffusion-xl-base-1.0
same vae license on sdxl-vae-fp16-fix
SDXL-VAE-FP16-Fix
SDXL-VAE-FP16-Fix is the SDXL VAE, but modified to run in fp16 precision without generating NaNs.
VAEDecoding in float32
/ bfloat16
precisionDecoding in float16
precisionSDXL-VAE✅⚠️SDXL-VAE-FP16-Fix✅✅
Details
SDXL-VAE generates NaNs in fp16 because the internal activation values are too big:
SDXL-VAE-FP16-Fix was created by finetuning the SDXL-VAE to:
1. keep the final output the same, but
2. make the internal activation values smaller, by
3. scaling down weights and biases within the network
There are slight discrepancies between the output of SDXL-VAE-FP16-Fix and SDXL-VAE, but the decoded images should be close enough for most purposes.
Benchmark from here:by Kubuxu
https://huggingface.co/madebyollin/sdxl-vae-fp16-fix/discussions/7
Evaluation on COCO val-2017, 256x256, RandomCrop with padding
Metrics:
LPIPS: https://github.com/richzhang/PerceptualSimilarity/ (lower better) and structural similarity index measure via skimage.metrics (higher better)
Metrics given as: mean [79% credibility interval]