Metraipolis › archive › dc_fd813ab5bbd5
Repairs for Whom? Evaluator-Executor Mismatch in Selecting Repairs for Failed LLM Agents — arXiv:2610.38610
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.38610
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-31T06:17:17.233Z
- SHA-256 of the archived text
1a837d37437b05585f42b7d987feab21683289d704551e67b5ffc6f6af287f0a- Stable URL of this page
https://metraipolis.com/documento/dc_fd813ab5bbd5
How to cite
Repairs for Whom? Evaluator-Executor Mismatch in Selecting Repairs for Failed LLM Agents — arXiv:2610.38610 — arXiv (en), https://arxiv.org/abs/2610.38610 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fd813ab5bbd5 (Metraipolis archive dc_fd813ab5bbd5, sha256 1a837d37437b05585f42b7d987feab21683289d704551e67b5ffc6f6af287f0a)
BibTeX
@misc{dcfd813ab5bbd5,
title = {Repairs for Whom? Evaluator-Executor Mismatch in Selecting Repairs for Failed LLM Agents — arXiv:2610.38610},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.38610},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fd813ab5bbd5; sha256 1a837d37437b05585f42b7d987feab21683289d704551e67b5ffc6f6af287f0a},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.38610. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
When an LLM agent fails, repair methods choose which action to replace and what to put in its place. They often score candidate repairs with greedy replay, a stronger model, an environment oracle, or a judge that reads the failed log. We ask whether these evaluators pick repairs that help the agent that will actually execute them. Holding failed trajectories and candidate single-step repairs fixed, we label every candidate with independent rollouts of the deployed policy and measure how much deployment value each evaluator's choice loses. On failures not screened for repair value, pre-specified tests show that greedy replay loses 0.08-0.13 in expected net success gain in ALFWorld and 0.06-0.13 in net reward in WebShop, relative to selection with the deployed agent's own rollouts. At an equal rollout budget, selecting repairs for Qwen2.5-7B with Qwen2.5-14B rollouts loses 0.25 in ALFWorld and 0.07 in WebShop, twenty 14B rollouts per candidate are worth no more than two of the 7B agent's own, and fresh rollouts confirm that the two models prefer opposite repairs on selected disagreement cases. Between similar-sized models from different families, the losses appear in both directions in both environments (0.07-0.17). The scoring executor also changes repair-location labels and method rankings. Repairs should be verified with the executor that will run them, and repair labels should name their executor. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno). Categorie arXiv: Software Engineering (cs.SE) ; Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2610.38610 — 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 2026-10-31T06:07:23.396Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).