Discofork
Repository brief

huggingface/diffusers

Read the upstream summary on the left, browse the cached forks below it, and load each fork comparison into the right-hand panel.

Cached analysis
cached 2026-03-30T15:55:07.106Z
1mo ago

huggingface/diffusers

huggingface/diffusers is an active, widely used Python library for pretrained diffusion models, focused on image, video, and audio generation. It is not archived, has a large community footprint, and is updated frequently.

GitHub
Loading tags...
Stars33,212
Forks6,884
Default branchmain
Last pushed2026-03-30T11:35:17Z
Recommended shortcuts

Jump straight into Discofork's strongest cached fork picks, or open a compare view in one click.

Forks

Choose a fork to inspect

10 of 10 fork briefs
Selected

Prefer this fork only if you need its older custom LoRA/attention-oriented behavior or legacy experiments. For most adopters, upstream is the safer choice because this fork is far behind, heavily diverged, and likely missing many recent fixes and features.

Prefer upstream unless you specifically need this fork's older Stability AI customizations. This fork looks like a legacy, highly divergent branch with useful niche model/workflow modifications but substantial missing upstream functionality and high maintenance risk.

Prefer this fork if your goal is FLUX/LoRA training and you value opinionated, model-specific workflows over general Diffusers coverage. Prefer upstream if you need broad model support, active maintenance, and the least-fragile base for long-term adoption.

Choose this fork only if you specifically want its video/pseudo-3D experimentation baseline and are comfortable owning compatibility work. If you want a maintained, broadly useful diffusion library, upstream Diffusers is the better choice.

Prefer this fork only if AMD/ONNX deployment is the goal and you are prepared to own a frozen, heavily diverged codebase. For most users, upstream Diffusers is the safer and more capable choice.

Choose this fork only if you specifically need its Flux/FLUX2 and LoRA-oriented customizations. If you want a maintained, broad-purpose diffusion library with the latest upstream fixes and examples, upstream Diffusers is the safer choice.

Prefer this fork only if you need its old custom code or historical snapshot. If you want an actively maintained diffusion framework, upstream is the better choice by a wide margin.

Prefer this fork only if its training-centric customizations are the point. If you want a current, broadly supported Diffusers base, upstream is the safer choice; this fork looks specialized, older, and materially harder to rebase.

Choose this fork only if you need its specific research or internal workflow customizations and are prepared to maintain a long-lived divergence. For most adopters, upstream diffusers is the safer choice because this fork is materially stale and missing substantial modern functionality.

Prefer this fork only if the goal is explicitly to remove upstream safety constraints and you are willing to own substantial maintenance debt. For most adopters, upstream Diffusers is the safer and more current choice.