[Bug]: Inpainting models fail with nonzero NGMS

[thot-experiment]

Is there an existing issue for this?

• [X] I have searched the existing issues and checked the recent builds/commits

What happened?

When attempting to use an inpainting model with nonzero NGMS the following error is raised:

RuntimeError: Sizes of tensors must match except in dimension 1. Expected size 1 but got size 2 for tensor number 1 in the list.

Steps to reproduce the problem

Use an inpainting model with non-zero negative guidance minimum sigma

What should have happened?

It should either work, or NGMS should be disabled with inpainting models.

Commit where the problem happens

20ae71f

What Python version are you running on ?

Python 3.10.x

What platforms do you use to access the UI ?

Windows, Android

What device are you running WebUI on?

Nvidia GPUs (GTX 16 below), Other GPUs

What browsers do you use to access the UI ?

Mozilla Firefox, Google Chrome

Command Line Arguments

--listen

List of extensions

all disabled

Console logs

venv "sdwui\venv\Scripts\Python.exe"
Python 3.10.7 (tags/v3.10.7:6cc6b13, Sep  5 2022, 14:08:36) [MSC v.1933 64 bit (AMD64)]
Version: v1.3.0
Commit hash: 20ae71faa8ef035c31aa3a410b707d792c8203a3
Installing requirements

Launching Web UI with arguments: --api --listen --enable-insecure-extension-access --xformers --port 7869 --cors-allow-origins *
ControlNet v1.1.200
ControlNet v1.1.200
importing, a thing that *should* work
and it did work
Loading weights [ec6f68ea63] from sdwui\models\Stable-diffusion\generic\lyriel_v16.safetensors
Create LRU cache (max_size=16) for preprocessor results.
Create LRU cache (max_size=16) for preprocessor results.
Creating model from config: sdwui\configs\v1-inference.yaml
LatentDiffusion: Running in eps-prediction mode
*Deforum ControlNet support: enabled*
DiffusionWrapper has 859.52 M params.
Loading VAE weights specified in settings: sdwui\models\VAE\vae-ft-ema-560000-ema-pruned.ckpt
Applying optimization: xformers... done.

Additional information

No response

[ktee95]

same problem

[Inblac]

same problem