diffusers

diffusers copied to clipboard

Add Ascend NPU support for SDXL and fix bugs

1

# What does this PR do? 1. Adds NPU flash attention support for NPU, similar to #7816. 2. Fixes a bug related to saving the model when using deepspeed, also...

HelloWorldBeginner

new from_single_file implementation is always internet-first and using local files only on timeout

12

### Describe the bug new `from_single_file` implementation is always internet-first and using local config files only after timeout even if they are already present locally as they were previously downloaded....

vladmandic

Add aspect ratio bucketing to training scripts

6

**Is your feature request related to a problem? Please describe.** When fine tuning SDXL, images are required to be a fixed size (1024x1024) which involves a lot of cropping that...

jferments

device_map='balanced' pipe.load_textual_inversion() function error

2

hi https://github.com/huggingface/diffusers/blob/b934215d4c376ea2e08e28103443686b95ea772c/src/diffusers/loaders/textual_inversion.py#L458 The code at this reference does not seem to handle device_map='balanced' and will report an error error msg: ``` diffusers\src\diffusers\pipelines\pipeline_utils.py", line 1130, in enable_sequential_cpu_offload raise ValueError( ValueError: It...

Honey-666

pipe.scheduler.sigmas change after forward call

5

### Describe the bug I noticed this issue when trying to manually modify the sigmas in my pipeline scheduler. Despite changing the values in this tensor, the sigmas tensor shrinks...

neuron-party

# What does this PR do? adds custom sigmas and timesteps for StableDiffusionXLControlNet pipeline; users can specify these two arguments in the forward call in accordance with https://github.com/huggingface/diffusers/pull/7817. ## Before...

neuron-party

Missing methods of `AutoencoderKL` in online docs

8

The [`decode`](https://github.com/huggingface/diffusers/blob/a080f0d3a21104c19effb4d9d95dcb710d99c19b/src/diffusers/models/autoencoders/autoencoder_kl.py#L281) method of `AutoencoderKL` is absent from the [online doc](https://huggingface.co/docs/diffusers/main/en/api/models/autoencoderkl#autoencoderkl); same for `encode`. Maybe docgen is affected by `apply_forward_hook`?

thethomasboyer

[Dreambooth] number of channels error in train_dreambooth.py

17

### Describe the bug I am trying to use train_dreambooth.py to train a personalized model by following https://github.com/huggingface/diffusers/tree/main/examples/dreambooth. I got the following error: ``` File "C:\WPy64-31090\python-3.10.9.amd64\lib\site-packages\torch\nn\modules\conv.py", line 458, in _conv_forward...

bssrdf

# What does this PR do? Fixes # (issue) ## Before submitting - [x] This PR fixes a typo or improves the docs (you can dismiss the other checks if...

liang-hou

`StableDiffusionXLInstructPix2PixPipeline` doesn't work with cosxl_edit

14

### Describe the bug CosXL Edit is an InstructPix2Pix model (https://huggingface.co/stabilityai/cosxl) released together with CosXL, however trying to load it gives a size mismatch error ### Reproduction ```py import torch...

apolinario