MemRL Model Card
这是可索引项目报告证据页：它保留 MemRL Model Card 的源材料入口、机制线索和限制提醒；正文仍需 reader/editor 与 academic public-copy review 后才能当作最终结论引用。
MemRL Model Card
| Field | Value |
|---|---|
| Repository | MemTensor/MemRL |
| Category | Runtime Reinforcement Memory |
| Stars / forks snapshot | 117 / 10 |
| Language | Python |
| License | MIT |
| Raw capture | raw-github/memtensor_memrl.md |
| Updated by | hourly public metadata update, 2026-05-24 |
1. Role in Self Evolve
MemRL 是自进化 agent 的论文代码，用 episodic memory 上的 runtime reinforcement learning 代替参数微调，通过环境反馈筛选高效策略并在多个 benchmark 上验证持续改进。
2. Working Principle
episodic memory -> two-phase retrieval -> environmental feedback -> runtime reinforcement update -> benchmark transfer
3. Evidence Path
existing raw capture promoted; web GitHub page observed 76 commits, MIT license, Python primary language, HLE/BigCodeBench/ALFWorld/Lifelong Agent Bench runners, 117 stars and 10 forks. Shell GitHub API access remained blocked by DNS and local gh auth was invalid, so this card treats the current snapshot as web-observed rather than API-verified.
4. Teaching Use
Use this card to explain Runtime Reinforcement Memory in the raw -> classification -> project card -> site/report pipeline. The reading path is: raw capture -> classification row -> public site card -> project report -> aggregate GitHub analysis.
5. Limits
当前未克隆源码，未运行 benchmark、SDK examples、skill install flows 或 memory experiments；star/fork/commit 快照来自公开 GitHub 页面文本或可见页面片段。