lllyasviel / lllyasviel/stable-diffusion-webui-forge

This may be my own system issue. But using a full vae causes the Kmodel to reload. Taesd doesn't. It often wastes a ton of time rendering.

Open
#1,282 0 comments 1 reaction 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
13k
Forks
1.7k
PR merge metrics
No merged PRs in 30d

Description

Perhaps its related to the issue people are having will the constant model reloads but I have noticed this completely goes away when set to TAESD, I get its faster and more lightweight but is that the only reason it's causing the models to reload. Is there something that can be done? If it's a bug that would be good since it could be fixed. Perhaps there's an optimization I can use?

o load target model IntegratedAutoencoderKL0

Contributor guide

No contributing guide indexed for this repository

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by reproducing the full-VAE and TAESD cases using the logged KModel and IntegratedAutoencoderKL load/unload sequence on CUDA. Trace when the VAE causes KModel to reload and compare it with the TAESD path. Done means full-VAE rendering no longer causes unnecessary repeated KModel reloads, or the reason and an applicable optimization are documented.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning, performance
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
28/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.