Metraipolis › archive › dc_bcb627201084
Taking the PULSE of Medical Reasoning: Step-Level Retrieval Timing for Knowledge-Grounded Inference — arXiv:2610.20292
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.20292
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-15T20:35:38.562Z
- SHA-256 of the archived text
7510ea17a132fbe586dc6322b9f8fb174286e1cc0935c39cf5beded390fe03a0- Stable URL of this page
https://metraipolis.com/documento/dc_bcb627201084
How to cite
Taking the PULSE of Medical Reasoning: Step-Level Retrieval Timing for Knowledge-Grounded Inference — arXiv:2610.20292 — arXiv (en), https://arxiv.org/abs/2610.20292 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bcb627201084 (Metraipolis archive dc_bcb627201084, sha256 7510ea17a132fbe586dc6322b9f8fb174286e1cc0935c39cf5beded390fe03a0)
BibTeX
@misc{dcbcb627201084,
title = {Taking the PULSE of Medical Reasoning: Step-Level Retrieval Timing for Knowledge-Grounded Inference — arXiv:2610.20292},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.20292},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bcb627201084; sha256 7510ea17a132fbe586dc6322b9f8fb174286e1cc0935c39cf5beded390fe03a0},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.20292. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Reliable medical reasoning requires LLMs to dynamically access external biomedical evidence to improve factual trustworthiness and reasoning accuracy. However, indiscriminate retrieval can introduce noise and latency, raising a critical retrieval-timing problem: at each reasoning step, the system must decide whether the model's internal knowledge is reliable or external evidence is needed. We propose PUlSE, a lightweight probing method that uses the model's internal representations as intrinsic signals for step-level retrieval control. Our posterior trajectory analysis reveals that erroneous steps exhibit substantially higher inter-layer volatility than correct ones, suggesting unstable internal knowledge. Based on this finding, PUlSE trains a compact probe on inter-layer geometric features to detect factual errors upon step completion, and uses it as an adaptive trigger for retrieval and step regeneration. Across four medical benchmarks, PUlSE-timed retrieval matches exhaustive per-step retrieval in accuracy while reducing retrieval calls by 72%, and can transfer to an unseen legal domain, indicating potential domain-agnostic generality. Overall, our study provides an initial step toward shifting retrieval-trigger design from surface-level confidence estimates to internal activation dynamics for adaptive access. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno). Categorie arXiv: Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2610.20292 — 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 2026-10-15T19:36:18.252Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).