Pinned Loading

1. hermes-workspace hermes-workspace Public

   Does your AI agent actually follow rules? 13 pre-registered experiments + 5-layer verification architecture. Paper, data, code — all public.

   Python

2. paper-validator paper-validator Public

   One-command audit: python -m paper_validator claim --claim all --trials 30. Validates 8 governance claims against live LLM APIs.

   Python

3. deepseek-claude-code-starter deepseek-claude-code-starter Public

   DeepSeek V4 inside Claude Code — the scaffold that started my AI systems work. 9 agents, security hooks, MIT. 2 stars.

   Python 2

4. compact-counter compact-counter Public

   LLM 上下文压缩 sweet-spot 检测工具 | Python标准库 | 发现过早压缩反而改善输出 | Part 3

5. digital-twin-trainer digital-twin-trainer Public

   核心项目 | QLoRA 行为内化实验 | loss下降但行为崩塌 — 微调需要行为级评测 | Part 2

   Python

6. training-gate training-gate Public

   工具 | 微调质量关卡 + 行为漂移检测（HF evaluate PR #778，pending review）

   Python