DeepInfer: High-performance, large-scale inference framework for DeepSeek models
DeepSeek Harness: Everything is a Plugin.
—
DeepGEMM: clean and efficient BLAS kernel library on GPU
FlashMLA: Efficient Multi-head Latent Attention Kernels
A lightweight library for xPU kernel JIT compilation
DeepEP: an efficient expert-parallel communication library
A kernel library written in tilelang
DeepGEMM-Ascend: clean and efficient matrix multiplication kernel library for Huawei Ascend NPUs
DeepSelect: TopK kernels for DeepSeek Sparse Attention (DSA) and Samplers
Codex 风格的推理等级滑条(DeepSeek Harness 插件):可拖动连续滑条 + 高档位渐变星尘特效
DeepSeek Harness (dsh) 从 0 到 1 深度手册:安装/插件开发/性能调优/实测案例/同模型多 Agent 实测对比(中文 + 英文 PDF)
A high-performance communication library for machine learning training and inference on Huawei Ascend NPUs.
An internal component used by DeepSeek Harness
Community Docker and Kubernetes packaging for DeepSeek Harness (@deepseek-ai/dsh), with a hardened image, Compose stack, Helm chart, Web UI, and headless CLI.
DeepSeek V4 Flash 0731 - free deepseek v4 flash download, gguf, ollama, deepseek v4 pro 0813. deepseek v4 flash vs pro, benchmark, MoE, 1M context, compressed sparse attention, coding. CUDA/CPU, no API key, Windows local AI. Download:🡇
🐋 DeepSeek Harness 插件总目录 · The catalog of DSH plugins:1958 个仓库 / 1819 个真插件(Skills · MCP · Tools · UI · Orchestration),中英文搜索、分类筛选、STAR 排序 → leenkcool.github.io
Run @deepseek-ai/dsh (DeepSeek Harness) on Android / Termux — 在Android/Termux上运行DeepSeek Harness
DeepSeek CLI
This Codex skill is adapted from DeepSeek Harness’s [dsh-find-simplifications](https://github.com/deepseek-ai/deepseek-harness/tree/master/.agents/skills/dsh-find-simplifications) skill.