diffusers

diffusers copied to clipboard

Ptxla updates

4

# What does this PR do? Fixes step time calculations for ptxla training script. ## Before submitting - [ ] This PR fixes a typo or improves the docs (you...

entrpn

Fix some documentation in ./src/diffusers/models/embeddings.py for demo

4

# What does this PR do? Fixes #9567 ## Before submitting - [x] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's...

DTG2005

ValueError: --image_column' value 'image' needs to be one of: text

13

### Describe the bug env: MODEL_NAME=runwayml/stable-diffusion-v1-5 env: INSTANCE_DIR=/content/drive/MyDrive/Newfolder env: HF_ENDPOINT=https://hf-mirror.com/ 2024-08-18 08:46:08.308678: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:485] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already...

SoumyaMB10

Redundant reinitialization of text encoders in train_dreambooth_lora_flux

2

### Describe the bug In the `train_dreambooth_lora_flux.py` script, during each call to `log_validation`, the text encoders `text_encoder_one` and `text_encoder_two` are reinitialized. https://github.com/huggingface/diffusers/blob/8ba90aa706a733f45d83508a5b221da3c59fe4cd/examples/dreambooth/train_dreambooth_lora_flux.py#L1768 This occurs even when the text encoders do...

CyberDragon93

Flux inversion

3

current img2img is not so well, [RF Inversion](https://rf-inversion.github.io/)) provides an inverse method for Flux real image editing, can we implement it using diffusers? or how can we use DDIM inversion...

yuxu915

Example run: https://github.com/huggingface/diffusers/actions/runs/11384275265

sayakpaul

Fix Attention Mask Padding to Ensure Multiple of 8 Alignment

2

# What does this PR do? Fixes #9637 resolve Attention Mask Padding Issue for Compatibility with xFormers ## Before submitting - [ ] This PR fixes a typo or improves...

SahilCarterr

Unable to Retrieve Intermediate Gradients with CogVideoXPipeline

3

### Describe the bug When generating videos using the CogVideoXPipeline model, we need to access the gradients of intermediate tensors. However, we do not require additional training or parameter updates...

lovelyczli

FIFO-Diffusion: Generating Infinite Videos from Text without Training through Rolling Video Denoising

3

### Model/Pipeline/Scheduler description The authors propose a novel inference technique based on a pretrained diffusion model for text-conditional video generation. Their approach, called FIFO-Diffusion, is conceptually capable of generating infinitely...

clarencechen

Confusion about `FrozenDict` in `configuration_utils.py`

2

I am confused about the design of `FrozenDict` in `configuration_utils.py` and the usage of it. ### 1. Is `FrozenDict` really frozen? From the code, `FrozenDict` sets `self.__frozen = True` during...

townwish4git