Metraipolis › archive › dc_7ad28385f67b
PAST-Bench: Benchmarking the Foundations of Recursive Self-Improvement in Personal Agents — arXiv:2608.04003
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.04003
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-31T12:31:19.032Z
- SHA-256 of the archived text
49296b15eb2edceb0df9e85f073f5f0a9787d56bf98ec084d3e228585450c216- Stable URL of this page
https://metraipolis.com/documento/dc_7ad28385f67b
How to cite
PAST-Bench: Benchmarking the Foundations of Recursive Self-Improvement in Personal Agents — arXiv:2608.04003 — arXiv (en), https://arxiv.org/abs/2608.04003 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7ad28385f67b (Metraipolis archive dc_7ad28385f67b, sha256 49296b15eb2edceb0df9e85f073f5f0a9787d56bf98ec084d3e228585450c216)
BibTeX
@misc{dc7ad28385f67b,
title = {PAST-Bench: Benchmarking the Foundations of Recursive Self-Improvement in Personal Agents — arXiv:2608.04003},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.04003},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7ad28385f67b; sha256 49296b15eb2edceb0df9e85f073f5f0a9787d56bf98ec084d3e228585450c216},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.04003. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Recursive self-improvement requires agents to turn accumulated experience into better future behavior. Personal AI agents offer a concrete setting for studying this capability because they retain preferences, task histories, tool routines, and learned skills across sessions. Yet whether retained experience actually improves them over time has not been systematically tested. We introduce PAST-Bench, a benchmark designed to isolate this question. Each agent runs through ordered sequences of fresh-session tasks under matched conditions that turn retained experience on and off. It spans 26 scenarios and 204 episodes across memory, procedural reuse, information gathering, and update. We report both later-task gains and whether those gains follow the intended save, retrieve, and update pathway. Across seven base models and four agent frameworks, improvement is real but uneven across capabilities. Agents with the same headline gain can differ markedly in whether that gain is supported by evidence of the intended pathway. Guided by these findings, we develop Hermes+, which extends Hermes with five targeted interventions across stages of the agent loop. Hermes+ raises the average gain from retained experience and provides clearer pathway evidence, with its strongest improvement on tasks requiring outdated state to be replaced, although the effect remains capability- and model-dependent. Together, PAST-Bench and Hermes+ provide an evaluation and diagnostic foundation for studying how persistent agents can progress from retaining experience to systematically improving through it. Code: this https URL Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-05 in cs.CL). Categorie arXiv: Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2608.04003 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-31T12:17:37.790Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-08-05?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).