Summary of Compass Memory Layer
Compass is an open-source memory layer designed to enhance AI agents' long-term memory capabilities. It aims to provide a robust and scalable solution for managing and retrieving contextual information over extended periods, addressing the limitations of existing systems like MemoBase, Mem0, Zep, and EverCore.
Key Features:
- Multi-party QA Support: Handles complex multi-party dialogues effectively.
- Cross-judge Replication: Ensures consistency across different evaluators with a Cohen's κ score of 0.772.
- Stress Benchmarking: Demonstrates high performance under heavy load, with p95 latency at 7ms for up to 1M rows in SQLite.
Performance Metrics:
- EverMemBench-Dynamic (2400 QAs):
- BM25 Baseline: R@1 = 14.8%, R@5 = 25.2%, R@10 = 30.6%, R@20 = 38.1%
- Compass Full Stack:
- Recall@20: 94.8%
- Accuracy: 41
Read the full article at DEV Community
Want to create content about this topic? Use Nemati AI tools to generate articles, social posts, and more.
24
Tags
Contents
Ali NematiWritten by Ali
