fastllm Agent Skills & AI Rules

GitHub

Fastllm是后端无依赖的高性能大模型推理库。同时支持张量并行推理稠密模型和混合模式推理MOE模型,任意10G以上显卡即可推理满血DeepSeek。双路9004/9005服务器+单显卡部署DeepSeek满血满精度原版模型,单并发20tps;INT4量化模型单并发30tps,多并发可达60+。

5,112 stars C++ C++ 104 files · ~126,163 tokens
RAW Rules
3 Discovered Rules • ~900 Estimated Tokens
Loading skill content...

Claude Plugin Marketplace Catalog

.claude-plugin/marketplace.json marketplace ~300 tokens
Raw
# Claude Plugin Marketplace Catalog

Path: `.claude-plugin/marketplace.json`

[View Raw Content on GitHub](https://raw.githubusercontent.com/ztxz16/fastllm/HEAD/.claude-plugin/marketplace.json)