PAST-Bench: Benchmarking the Foundations of Recursive Self-Improvement in Personal Agents
- 类型:arxiv
- 标识:2608.04003
- 链接:http://arxiv.org/abs/2608.04003v1
- 主分类:agent
- 形态:benchmark
- 被引:0
- 被引来源:Semantic Scholar
- S2被引:0
- 影响力被引:0
- TLDR:The PAST-Bench benchmark is introduced, a benchmark designed to isolate how persistent agents can progress from retaining experience to systematically improving through it, and Hermes+ is developed, which raises the average gain from retained experience and provides clearer pathway evidence.
- 副分类:evaluation
- 待LLM分类:否
- 标题中文:PAST-Bench:面向个人 Agent 递归自我改进基础的基准评测
- TLDR中文:提出 PAST-Bench 基准,用于隔离评估持久 agent 从经验保留到系统性改进的能力;开发 Hermes+,提升了来自保留经验的平均增益,并提供更清晰的路径证据。
- 来源文件:
- /inbox/tom/_candidates/2026-08-05-agent-rag-longcontext-candidates.json
- /inbox/tom/_candidates/2026-08-06-agent-rag-longcontext-candidates.json
- [S2 enrich]