diffusers

diffusers copied to clipboard

Bug of PixArtAlphaPipeline

1

### Describe the bug When I use with torch.autocast("cuda"): images = pipe(...) the output is noise  ### Reproduction import torch from diffusers import PixArtAlphaPipeline pipe = PixArtAlphaPipeline.from_pretrained("PixArt-alpha/PixArt-XL-2-1024-MS", torch_dtype=torch.float16) pipe.enable_model_cpu_offload()...

CuddleSabe

Mismatching size in matmul when using StableDiffusionInstructPix2Pix pipeline with IP-Adapter

3

### Describe the bug I've tried to combine InstructPix2Pix model and IP-Adapter (`pipe.load_ip_adapter("h94/IP-Adapter", subfolder="models", weight_name="ip-adapter_sd15.bin"`), but I always get `RuntimeError: mat1 and mat2 shapes cannot be multiplied (771x1280 and 1024x3072)`....

misshimichka

Controlnet inpainting change too much of masking area

5

### Describe the bug mask   i got mask with this ``` def make_inpaint_condition (image, image_mask): image = np.array(image.convert("RGB")).astype(np.float32) / 255.0 image_mask = np.array(image_mask.convert("L")).astype(np.float32) / 255.0 assert image.shape[0:1] ==...

crapthings

Check for latents, before calling prepare_latents - sdxlImg2Img

7

# What does this PR do? Fixes #7011 ## Before submitting - [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks if...

nileshkokane01

# What does this PR do? There are a couple of differences between transformers workflows and ours. e.g. We don't have a clone our repo in our docker image. Update...

DN6

fix multicontrolnet `save_pretrained` logic for compatibility

2

# What does this PR do? Fix an inconsistency between `from_pretrained` and `save_pretrained` within the Multicontrolnet class. The `from_pretrained` function returns a directory structure like this: controlnet, controlnet_1, controlnet_2, whereas...

rebel-kblee

Fix key error for dictionary with randomized order in convert_ldm_unet_checkpoint

6

# What does this PR do? I found that for some model weights, the key values in the dictionary are not applied in order, and this can cause a potential...

yunseongcho

Train GLIGEN in diffusers?

10

Thanks for your great job!!! Now, I know how to infer GLIGEN with diffusers[https://github.com/gligen/diffusers/tree/gligen/examples/gligen](url). But how can I train GLIGEN with diffusers like ControlNet[https://github.com/huggingface/diffusers/blob/main/examples/controlnet/train_controlnet.py](url)? Thanks again.

Strike1999

VAE training sample script

90

I believe the current lack of easy access to VAE training is stopping diffusion models from disrupting even more industries. I'm talking about consistent details on things that are less...

zhuliyi0

Add Ascend NPU support for SDXL fine-tuning and fix the model saving bug when using DeepSpeed.

11

# What does this PR do? Added support for SDXL finetune on AscendNPU and fixed the bug causing the hang out when saving models using the deepspeed distributed framework. DeepSpeed...

HelloWorldBeginner