Ryan Dick

DoRA loading does not load all keys from the state_dict

9

### Describe the bug When loading a DoRA model from a kohya state_dict some keys are silently skipped in the state_dict. DoRA loading was added in https://github.com/huggingface/diffusers/pull/7371. This feature has...

### Is there an existing issue for this problem? - [X] I have searched the existing issues ### Operating system Linux ### GPU vendor Nvidia (CUDA) ### GPU model _No...

## Summary Adds a `TiledStableDiffusionRefineInvocation` node with the following characteristics: - The node applies the following sequence: - Split tiles in image space - VAE encode - Denoise - VAE...

## Problem The InvokeAI import graph is a tangled web ripe with circular imports and import-order dependencies. We currently work around these issues with short-term hacks, but they are a...

## Summary This PR add support for FLUX LoRA models in kohya format with `lora_te1` layers (i.e. CLIP LoRA layers). Previously, only transformer LoRA layers were supported. Example LoRA model...

## Summary Add support for FLUX ControlNet models (XLabs and InstantX). ## QA Instructions - [ ] SD1 and SDXL ControlNets, since the ModelLoaderRegistry calls were changed. - [x] Single...

## Summary This PR adds a workaround for broken sliced attention on MPS with torch 2.4.1. The workaround keeps things working on MPS at the cost of increased peak memory...

Configs for reproducing style preference experiments

1

Thank you for open sourcing your work. I am trying to reproduce the style-preference training experiments from the paper (Cartoon and Pixel Art). Can you provide guidance on what configs...