huggingface/diffusers

View on GitHub

🤗 Diffusers: State-of-the-art diffusion models for image, video, and audio generation in PyTorch.

Stars: 34.5k
Forks: 7.3k
Open beginner issues: 22
Indexed issues: 838
Avg merge: 3d 3h
Merged PRs (30d): 91

Dominant language: Python
License: Apache-2.0
Last GitHub push: Sep 15, 2026
Latest indexed: Sep 17, 2026
Contributing guide: Contributing guide
Code of conduct: Code of conduct
Beginner labels: help wanted good first issue

Open All

838 open issues indexed

deepspeed train flux1 dreambooth lora can not save model Open

bug

huggingface/diffusers#9393 · 21 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
[Q] Possibly unused `self.final_alpha_cumprod` Open

stale

huggingface/diffusers#9395 · 7 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Suggestion for speeding up `index_for_timestep` by removing sequential `nonzero()` calls in samplers Open

contributions-welcome help wanted performance wip

huggingface/diffusers#9417 · 11 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Remove redundant comparison inside the diffusion loop of stable video diffusion pipeline Open

stale

huggingface/diffusers#9425 · 8 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Formulation of reverse diffusion process in DDPM ( 1 - alpha_prod_t = beta_prod_t assumption) Open

bug stale

huggingface/diffusers#9474 · 5 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
train_dreambooth_lora_flux validation RuntimeError: Input type (float) and bias type (c10::BFloat16) should be the same Open

bug

huggingface/diffusers#9476 · 11 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
FLUX dreambooth train on multigpu with deepspeed Open

bug stale

huggingface/diffusers#9484 · 13 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Can we allow making everything on gpu/cuda for scheduler? Open

performance scheduler stale

huggingface/diffusers#9485 · 14 comments · 0 reactions · 2 assignees · Indexed Sep 17, 2026
Dreambooth Flux training failed on saving a checkpoint Open

bug stale

huggingface/diffusers#9500 · 26 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Dreambooth Flux training does not save a model for around 10-15 minutes Open

bug stale

huggingface/diffusers#9501 · 16 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Confusion about `FrozenDict` in `configuration_utils.py` Open

stale

huggingface/diffusers#9503 · 3 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
AnimateDiff SparseCtrl RGB does not work as expected Open

advanced bug contributions-welcome help wanted stale

huggingface/diffusers#9508 · 9 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Multi-controlnet batching for `StableDiffusionXLControlNetInpaintPipeline` Open

contributions-welcome stale

huggingface/diffusers#9511 · 5 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Freeing GPU memory after `torch.compile` StableDiffusionXLPipeline UNet Open

stale torch-compile

huggingface/diffusers#9530 · 4 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Problems with saved lora models when using rslora. Open

bug stale

huggingface/diffusers#9556 · 7 comments · 2 reactions · 0 assignees · Indexed Sep 17, 2026
APG: Eliminating Oversaturation and Artifacts of High Guidance Scales in Diffusion Models Open

consider-for-modular-diffusers stale

huggingface/diffusers#9585 · 6 comments · 12 reactions · 1 assignee · Indexed Sep 17, 2026
FlowMatch schedulers - closing the gap Open

scheduler wip

huggingface/diffusers#9607 · 11 comments · 0 reactions · 1 assignee · Indexed Sep 17, 2026
process_encoder_hidden_states function is applied on each denoising step Open

stale

huggingface/diffusers#9625 · 6 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
[Schedulers] Add methods for conversion between noise and denoised predictions Open

scheduler stale

huggingface/diffusers#9629 · 5 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Error when setting num_single_layers=0 while training flux-controlnet on a multi-GPU server using a single GPU Open

bug stale

huggingface/diffusers#9630 · 8 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
[Flux ControlNet] Add support for de-distilled models with CFG Open

contributions-welcome good first issue

huggingface/diffusers#9635 · 15 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
XFormer fails when passing attention mask while using bfloat and key's sequence length not being a multiple of 8 Open

bug stale

huggingface/diffusers#9637 · 4 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Code With Strange Logic in CogVideoX's Dynamic CFG Open

bug wip

huggingface/diffusers#9641 · 3 comments · 3 reactions · 0 assignees · Indexed Sep 17, 2026
Error in loading flux fp8 model with local transformer_flux.py file Open

bug stale

huggingface/diffusers#9667 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
[SD3 ControlNet] bug in pipeline 'controlnet_pooled_projections' Open

bug stale

huggingface/diffusers#9686 · 11 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
FLUX.1-dev FP8 Example Code Cleanup Open

huggingface/diffusers#9690 · 8 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Ctrlora implementation : Easy solution to train a controlnet with < 1000 examples Open

contributions-welcome stale

huggingface/diffusers#9713 · 5 comments · 6 reactions · 0 assignees · Indexed Sep 17, 2026
How to use Playground2.5 to train lora with own dataset to generate pictures of a specific style？ Open

bug stale

huggingface/diffusers#9731 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Add Ultimate SD Upscale pipeline for high-quality tiled image upscaling Open

huggingface/diffusers#9740 · 22 comments · 0 reactions · 1 assignee · Indexed Sep 17, 2026
Is it possible to provide img2img code for CogView3? Open

contributions-welcome stale

huggingface/diffusers#9750 · 3 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Support SOTA Difussion Based talking head repos like Echomimic if not already supported Open

stale

huggingface/diffusers#9783 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Add support to Meissonic Open

contributions-welcome stale

huggingface/diffusers#9794 · 7 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Loading pipeline in precision it was saved in Open

enhancement

huggingface/diffusers#9797 · 10 comments · 3 reactions · 0 assignees · Indexed Sep 17, 2026
Multidiffusion (panorama pipeline) is missing segmentation inputs? Open

stale

huggingface/diffusers#9802 · 5 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Flux LoRA loading failure Open

bug

huggingface/diffusers#9803 · 7 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Lora does not support module_to_save() Open

bug stale

huggingface/diffusers#9810 · 2 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
`prepare_ip_adapter_image_embeds` Bug Causes Feature Mixing During Batch Processing in IP-Adapter Open

bug

huggingface/diffusers#9813 · 4 comments · 1 reaction · 2 assignees · Indexed Sep 17, 2026
ConnectionError: Tried to launch distributed communication on port 29401, but another process is utilizing it. Please specify a different port (such as using the --main_process_port flag or specifying a different main_process_port in your config file) and rerun your script. To automatically use the next open port (on a single node), you can set this to 0. Open

bug stale

huggingface/diffusers#9856 · 7 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Flux training seems not to update the transformer model Open

bug stale

huggingface/diffusers#9861 · 7 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Why isn’t VRAM being released after training LoRA? Open

bug stale

huggingface/diffusers#9876 · 14 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
PAG for StableDiffusionControlNetImg2ImgPipeline Open

contributions-welcome stale

huggingface/diffusers#9886 · 6 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
lack of support of loading lora weights in PixArtAlphaPipeline Open

contributions-welcome lora

huggingface/diffusers#9887 · 8 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Segmentation fault (core dumped) when sdxl-turbo inference with torch 2.2.1+cu118 Open

bug stale

huggingface/diffusers#9889 · 4 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
UNet2DConditionModel to onnx with torch.onnx faild Open

bug stale

huggingface/diffusers#9890 · 4 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
multi controlnet error for flux when using 2 controlnet with different layer length Open

bug wip

huggingface/diffusers#9911 · 6 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Why set_lora_device doesn't work Open

bug needs-code-example stale

huggingface/diffusers#9913 · 8 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
[Community] Add MagicTailor Personalization Training Script Open

contributions-welcome stale

huggingface/diffusers#9923 · 1 comment · 0 reactions · 0 assignees · Indexed Sep 17, 2026
Can we get more schedulers for flow based models such as SD3, SD3.5, and flux Open

scheduler wip

huggingface/diffusers#9924 · 40 comments · 0 reactions · 0 assignees · Indexed Sep 17, 2026
[PAG] - Adaptive Scale bug Open

bug stale

huggingface/diffusers#9930 · 1 comment · 1 reaction · 0 assignees · Indexed Sep 17, 2026
Improve SD35 LoRA support to cover most popular LoRA formats Open

huggingface/diffusers#9950 · 9 comments · 1 reaction · 1 assignee · Indexed Sep 17, 2026

huggingface/diffusers

Get new issues in your inbox