Metraipolis › archive › dc_0ee9baebfb61

S3Gym: Can LLMs Turn Self-Testing and Self-Judging into Self-Improvement? — arXiv:2608.31100

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.31100
License (as archived)
CC0-1.0
Archived
2026-10-17T08:07:55.381Z
SHA-256 of the archived text
1dcd7d0ff52a03a43e211ae7b9145a7c7c52d1aa37bdf2fd3c85cd1cd9c5e8d2
Stable URL of this page
https://metraipolis.com/documento/dc_0ee9baebfb61

How to cite

S3Gym: Can LLMs Turn Self-Testing and Self-Judging into Self-Improvement? — arXiv:2608.31100 — arXiv (en), https://arxiv.org/abs/2608.31100 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0ee9baebfb61 (Metraipolis archive dc_0ee9baebfb61, sha256 1dcd7d0ff52a03a43e211ae7b9145a7c7c52d1aa37bdf2fd3c85cd1cd9c5e8d2)
BibTeX
@misc{dc0ee9baebfb61,
  title = {S3Gym: Can LLMs Turn Self-Testing and Self-Judging into Self-Improvement? — arXiv:2608.31100},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.31100},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0ee9baebfb61; sha256 1dcd7d0ff52a03a43e211ae7b9145a7c7c52d1aa37bdf2fd3c85cd1cd9c5e8d2},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.31100. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large language models (LLMs) increasingly interact with external environments and accumulate substantial behavioral experience, yet existing agent benchmarks largely evaluate them as fixed policies. It therefore remains unclear whether an agent can actively test its behavior, judge the resulting experience, and use that experience to improve future decisions. We introduce \textbf{S\textsuperscript{3}Gym}, an interactive benchmark for evaluating LLM self-improvement through three coupled capabilities: \textbf{Self-Testing}, \textbf{Self-Judging}, and \textbf{Self-Improvement}. S$^3$Gym separates permissive exploration from strict held-out evaluation and instantiates this protocol in seven text-based games with executable environment verifiers. We evaluate three pathways for incorporating interaction experience: direct History ICL, score-conditioned Summary Memory, and parameter Training. Our experiments reveal that self-improvement is neither automatic nor uniform. Context-level experience improves performance for several model--game pairs, but the most effective pathway depends strongly on the task structure: summaries are beneficial when experience can be compressed into reusable strategic rules, yet often underperform raw history when success depends on precise, state-contingent information. Parameter training produces substantial gains on some tasks, but also exhibits unstable improvement and severe negative transfer on others. These findings show that recognizing successful actions is insufficient; agents must also transform feedback into executable and transferable policies. S$^3$Gym provides a unified framework for diagnosing this process and identifying the bottlenecks that prevent agents from translating interaction experience into reliable self-improvement.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2608.31100 — 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 2026-10-17T07:47:52.885Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).