vipshop/cache-dit

A PyTorch-native inference engine with cache, parallelism, quantization and cpu offload for 100+ DiTs.

About the project

Cache-DiT is a PyTorch inference engine for DiT models, with caching, parallel processing, quantization, and CPU offloading. It is for running diffusion-model pipelines with support for systems and hardware such as Diffusers, SGLang Diffusion, vLLM-Omni, ComfyUI, NVIDIA GPUs, Ascend NPUs, and AMD GPUs.

Latest release v1.5.3 · 15 Oct 2026

4 files for Claude Code: cache-dit-model-integration, cache-dit-build-release, cache-dit-custom-kernel, cache-dit-quantization .

Open on GitHub cache-dit.readthedocs.io

Skills

cache-dit-model-integration .agents/skills/cache-dit-model-integration/SKILL.md A 243 tok
cache-dit-build-release .agents/skills/cache-dit-build-release/SKILL.md A 191 tok
cache-dit-custom-kernel .agents/skills/cache-dit-custom-kernel/SKILL.md A 136 tok
cache-dit-quantization .agents/skills/cache-dit-quantization/SKILL.md A 106 tok

These files are vipshop/cache-dit's own configuration — they tell Claude Code how to work on this repository, so they are not mods to install elsewhere. Copy one as a starting point and replace the parts that are about this project.