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 18, 2026
Latest indexed: Sep 17, 2026
Contributing guide: Contributing guide
Code of conduct: Code of conduct
Beginner labels: help wanted good first issue

Beginner friendly Everything

Open All

838 issues indexed so far

53 messages in src/diffusers are missing the f-string prefix (safety-checker ValueError in 46 pipelines) Open

Difficulty 3/5 1-2 days Newbie friendliness 68/100

huggingface/diffusers#14803 · Opened Sep 17, 2026
Revisit modular diffusers weight-duplication workaround now that huggingface_hub has a shared blobs cache Open

huggingface/diffusers#14802 · 2 assignees · Opened Sep 17, 2026
TangentialClassifierFreeGuidance.is_conditional reads an attribute that does not exist Open

bug needs-env-info pipelines

Difficulty 2/5 1-3 hours Newbie friendliness 72/100

huggingface/diffusers#14794 · Opened Sep 16, 2026
Opt-in weight integrity verification at load time (pinned SHA-256 hashes / manifest) Open

feature-request pipelines

Difficulty 4/5 3-5 days Newbie friendliness 48/100

huggingface/diffusers#14784 · Opened Sep 15, 2026
Deprecating `Siglip2ImageProcessorFast` Open

bug needs-code-example needs-env-info pipelines

Difficulty 1/5 Under an hour Newbie friendliness 86/100

huggingface/diffusers#14780 · 1 comment · Opened Sep 15, 2026
MLU backend utilities fail despite MLU device detection Open

bug models needs-env-info

Difficulty 3/5 1-2 days Newbie friendliness 74/100

huggingface/diffusers#14779 · 5 comments · Opened Sep 15, 2026
LTX2 import fails with Transformers 5.5.0 due to annotation-only Gemma4Unified dependency Open

bug needs-env-info pipelines

Difficulty 3/5 1-2 days Newbie friendliness 76/100

huggingface/diffusers#14773 · 2 comments · Opened Sep 14, 2026
[Bug / Optimization] Make AutoencoderKLWan stateless to fix torch.compile compatibility and enable 1.45x speedup Open

bug models

Difficulty 4/5 3-5 days Newbie friendliness 68/100

huggingface/diffusers#14770 · Opened Sep 14, 2026
save_pretrained(safe_serialization=False) leaves the old safetensors checkpoint behind, and from_pretrained loads it instead of the new weights Open

bug pipelines

Difficulty 2/5 1-3 hours Newbie friendliness 78/100

huggingface/diffusers#14769 · 1 comment · Opened Sep 14, 2026
`Cosmos2VideoToWorldPipeline` applies FPS RoPE modulation that the released Cosmos-Predict2 checkpoints disable natively (default fps=16 → temporal RoPE ×1.5) Open

bug models needs-env-info pipelines

Difficulty 4/5 3-5 days Newbie friendliness 64/100

huggingface/diffusers#14768 · 1 comment · Opened Sep 14, 2026
Add IP-Adapter support to StableDiffusionXLInstructPix2PixPipeline Open

feature-request pipelines

Difficulty 4/5 3-5 days Newbie friendliness 38/100

huggingface/diffusers#14761 · 2 comments · Opened Sep 12, 2026
Pipeline properties that read attributes __call__ never sets (SD3 skip_guidance_layers, LEdits++ XL guidance_scale) Open

bug needs-env-info pipelines

Difficulty 4/5 3-5 days Newbie friendliness 38/100

huggingface/diffusers#14759 · Opened Sep 12, 2026
move `LTX2TextConnectors` into `models/condition_embedders/` Open

Difficulty 3/5 1-2 days Newbie friendliness 76/100

huggingface/diffusers#14749 · 1 comment · Opened Sep 9, 2026
MiniMax-H3 VAE decoder loads in fp32 but CUDA decode runs it in fp16 Open

bug modular-pipelines needs-env-info

Difficulty 4/5 3-5 days Newbie friendliness 55/100

huggingface/diffusers#14746 · 1 comment · Opened Sep 9, 2026
[Community] Rectified-CFG++ pipelines for SD3/3.5, Flux, Lumina 2, Qwen-Image, and Wan Open

feature-request pipelines

Difficulty 5/5 Over a week Newbie friendliness 20/100

huggingface/diffusers#14744 · Opened Sep 9, 2026
[modular] move unpack out of the decoder blocks for flux, krea2 and ltx Open

contributions-welcome

Difficulty 4/5 3-5 days Newbie friendliness 68/100

huggingface/diffusers#14730 · 3 comments · Opened Sep 7, 2026
AttentionModuleMixin.set_attention_slice references two attributes that do not exist Open

bug models needs-env-info

Difficulty 5/5 Over a week Newbie friendliness 35/100

huggingface/diffusers#14729 · Opened Sep 7, 2026
Three docstrings type a boolean parameter as str Open

bug modular-pipelines needs-env-info pipelines

Difficulty 1/5 1-3 hours Newbie friendliness 25/100

huggingface/diffusers#14723 · Opened Sep 7, 2026
Community pipeline: rerank Stable Diffusion candidates by CLIP score Open

feature-request pipelines

Difficulty 4/5 3-5 days Newbie friendliness 55/100

huggingface/diffusers#14721 · Opened Sep 6, 2026
save_pretrained can delete shards from another variant and leave stale indexes Open

bug models needs-env-info

Difficulty 3/5 1-2 days Newbie friendliness 76/100

huggingface/diffusers#14719 · 1 comment · Opened Sep 6, 2026
[zh] Translating docs to Simplified Chinese Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

huggingface/diffusers#14711 · Opened Sep 5, 2026
[Quantization] Add support for Comfy Quants backend Open

huggingface/diffusers#14705 · 3 comments · 1 assignee · Opened Sep 4, 2026
Add VideoDeltaNet-H3 Model Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

huggingface/diffusers#14700 · 13 comments · Opened Sep 4, 2026
Fix skipped `test_model_xattn_padding` test by updating mask padding logic Open

bug models needs-code-example needs-env-info

Difficulty 4/5 3-5 days Newbie friendliness 55/100

huggingface/diffusers#14699 · 2 comments · Opened Sep 3, 2026
GGUF: enable_sequential_cpu_offload() raises KeyError: None — any GGUF model larger than VRAM is unrunnable Open

bug gguf models needs-env-info quantization

huggingface/diffusers#14691 · 3 comments · 1 assignee · Opened Sep 2, 2026
Missing-dependency error suggests `pip install pytorch_retinaface`, which is not a PyPI package Open

Difficulty 3/5 1-2 days Newbie friendliness 68/100

huggingface/diffusers#14666 · 2 comments · Opened Aug 31, 2026
`guidance_scale=1.0` no longer disables CFG in LTX-2 pipelines Open

bug

Difficulty 4/5 3-5 days Newbie friendliness 68/100

huggingface/diffusers#14649 · Opened Aug 30, 2026
Proposal: tutorial for synchronized ControlNet pair augmentation with AlbumentationsX Open

Difficulty 3/5 1-2 days Newbie friendliness 68/100

huggingface/diffusers#14644 · 1 comment · Opened Aug 28, 2026
MiniMax-H3: float64 rotary position grid cannot be moved to MPS (`Cannot convert a MPS Tensor to float64`) Open

bug

Difficulty 2/5 1-3 hours Newbie friendliness 88/100

huggingface/diffusers#14639 · Opened Aug 28, 2026
give state dict creation its own method in IP-adapter pipeline tests Open

huggingface/diffusers#14634 · 1 assignee · Opened Aug 28, 2026
Kolors text encoder carries an unreachable torch<2 attention path built on the MPS-unsafe baddbmm(empty, beta=0) idiom Open

Difficulty 2/5 1-3 hours Newbie friendliness 25/100

huggingface/diffusers#14624 · Opened Aug 27, 2026
document output_type="pt" in the agent docs Open

Difficulty 1/5 Under an hour Newbie friendliness 88/100

huggingface/diffusers#14616 · Opened Aug 27, 2026
Provenance/licensing clarification for controlnet-canny-sdxl-1.0 Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

huggingface/diffusers#14595 · Opened Aug 24, 2026
Update remaining pipeline docstrings from `torch_dtype` to `dtype` Open

Difficulty 2/5 1-3 hours Newbie friendliness 84/100

huggingface/diffusers#14591 · Opened Aug 24, 2026
RMSNorm crashes on NPU when elementwise_affine=False (weight=None): npu_rms_norm requires a real gamma tensor Open

bug

Difficulty 3/5 1-2 days Newbie friendliness 68/100

huggingface/diffusers#14590 · Opened Aug 24, 2026
UnCLIPScheduler.set_timesteps crashes (ZeroDivisionError) for 1 step and silently returns an empty schedule for 0 steps Open

Difficulty 2/5 1-3 hours Newbie friendliness 30/100

huggingface/diffusers#14586 · Opened Aug 24, 2026
fixed_large_log variance sampling returns NaNs Open

Difficulty 3/5 1-2 days Newbie friendliness 76/100

huggingface/diffusers#14569 · 1 comment · Opened Aug 23, 2026
train_lcm_distill_lora_sdxl.py: intermediate validation logs the un-adapted base model; saved LoRA mixes key conventions and drops all attention adapters on load Open

Difficulty 4/5 3-5 days Newbie friendliness 48/100

huggingface/diffusers#14548 · 2 comments · Opened Aug 21, 2026
better reporting of errors when using TP Open

huggingface/diffusers#14533 · 1 assignee · Opened Aug 19, 2026
The Diffusers MVP 🚀 — Ideas Track 💡 Open

diffusers-mvp

huggingface/diffusers#14529 · 6 comments · 1 reaction · 3 assignees · Opened Aug 18, 2026
Add support for JoyAI-Video-Edit Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

huggingface/diffusers#14524 · 1 comment · Opened Aug 18, 2026
Where should a training-free high-resolution hook for Flux (DyPE / spectral attention) live? Open

Difficulty 5/5 Over a week Newbie friendliness 30/100

huggingface/diffusers#14520 · 1 comment · Opened Aug 17, 2026
MiniMax Music 3 modular pipeline: 2.9x consumer-GPU speedup - findings that may be worth upstreaming Open

Difficulty 5/5 Over a week Newbie friendliness 30/100

huggingface/diffusers#14486 · 7 comments · 1 reaction · Opened Aug 15, 2026
[Community pipeline] Echo-Memory overlay for Wan 2.1 1.3B Open

feature-request pipelines

Difficulty 4/5 3-5 days Newbie friendliness 25/100

huggingface/diffusers#14482 · 2 comments · Opened Aug 15, 2026
Thoughts on maintaining our lightweight Docker container to reduce startup time for remote inference? Open

feature-request pipelines

Difficulty 5/5 Over a week Newbie friendliness 25/100

huggingface/diffusers#14473 · 1 comment · Opened Aug 14, 2026
InpaintProcessor.preprocess returns a bare tensor when no mask is passed, breaking its own 3-value contract Open

bug pipelines

Difficulty 2/5 1-3 hours Newbie friendliness 76/100

huggingface/diffusers#14470 · 1 comment · Opened Aug 13, 2026
[modular] `guider` should be an optional component for text encoder blocks Open

feature-request modular-pipelines

huggingface/diffusers#14469 · 1 assignee · Opened Aug 13, 2026
`FlowMatchEulerDiscreteScheduler.set_timesteps` silently ignores explicit `timesteps` since #14011 — intended semantics change or unreleased regression? Open

bug schedulers

Difficulty 4/5 3-5 days Newbie friendliness 45/100

huggingface/diffusers#14461 · 6 comments · Opened Aug 13, 2026
`register_to_config` mislabels positional `__init__` args as `_use_default_values`, so `from_config` round trips silently revert them to defaults Open

bug models

Difficulty 2/5 1-3 hours Newbie friendliness 78/100

huggingface/diffusers#14460 · 3 comments · Opened Aug 13, 2026
Modular pipelines: variant blocksets sharing a `model_name` init the wrong pipeline class Open

bug modular-pipelines needs-code-example needs-env-info

huggingface/diffusers#14451 · 1 assignee · Opened Aug 12, 2026

huggingface/diffusers

Get new issues in your inbox