Metraipolis › archive › dc_6b31852cd55d
What Should the Reflector See? An Empirical Study of Evidence in Reflective Prompt Optimization — arXiv:2609.32452
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.32452
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-07T21:50:08.438Z
- SHA-256 of the archived text
4a693d62f9bc96725a5319586c228172370956cfd6c12f1a48522948b163e866- Stable URL of this page
https://metraipolis.com/documento/dc_6b31852cd55d
How to cite
What Should the Reflector See? An Empirical Study of Evidence in Reflective Prompt Optimization — arXiv:2609.32452 — arXiv (en), https://arxiv.org/abs/2609.32452 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_6b31852cd55d (Metraipolis archive dc_6b31852cd55d, sha256 4a693d62f9bc96725a5319586c228172370956cfd6c12f1a48522948b163e866)
BibTeX
@misc{dc6b31852cd55d,
title = {What Should the Reflector See? An Empirical Study of Evidence in Reflective Prompt Optimization — arXiv:2609.32452},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.32452},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_6b31852cd55d; sha256 4a693d62f9bc96725a5319586c228172370956cfd6c12f1a48522948b163e866},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.32452. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Reflective prompt optimization revises instructions using examples of a model's behavior, but which evidence the reflector should receive remains unclear. We study evidence composition, visibility of examples, candidate selection and domain-knowledge policy within a single-parent Pareto-guided search. Using Qwen3.5-9B as both task model and reflector, we evaluate nine reflection strategies on five datasets. From these experiments, we find three distinct patterns. For performance improvement, Failures-only produces the largest mean test gain (+8.0 percentage points), while Balanced-mix and No-examples+Val share the best mean performance rank. For reflection effectiveness, No-examples achieves the best rank for improving sampled parents, yet yields only a 1.4-point mean test gain: local reflection success does not necessarily produce a stronger final prompt. For overfitting assessment, Failures-only and Balanced-mix share the lowest mean calibration-gap rank, while larger gaps on GPQA and IFBench show that calibration gains can overstate held-out improvement. This gap is a descriptive indicator, not a direct measure of overfitting. Together, these results show why reflection strategies should be assessed separately on final performance, parent improvement and calibration-to-test transfer. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.CL). Categorie arXiv: Computation and Language (cs.CL) ; Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2609.32452 — 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 2026-10-07T21:27:08.429Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).