Metraipolis › archive › dc_7a866aec1892

Efficient Reasoning Training Does Not Always Harm CoT Faithfulness and Monitorability — arXiv:2610.03509

Source
arXiv
Original
https://arxiv.org/abs/2610.03509
License (as archived)
CC0-1.0
Archived
2026-10-05T22:49:40.745Z
SHA-256 of the archived text
59e0c4987087feabaa73e89c64061e07218be3e7825c956e00dfa8c63d08fd2e
Stable URL of this page
https://metraipolis.com/documento/dc_7a866aec1892

How to cite

Efficient Reasoning Training Does Not Always Harm CoT Faithfulness and Monitorability — arXiv:2610.03509 — arXiv, https://arxiv.org/abs/2610.03509 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_7a866aec1892 (Metraipolis archive dc_7a866aec1892, sha256 59e0c4987087feabaa73e89c64061e07218be3e7825c956e00dfa8c63d08fd2e)
BibTeX
@misc{dc7a866aec1892,
  title = {Efficient Reasoning Training Does Not Always Harm CoT Faithfulness and Monitorability — arXiv:2610.03509},
  howpublished = {arXiv},
  url = {https://arxiv.org/abs/2610.03509},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_7a866aec1892; sha256 59e0c4987087feabaa73e89c64061e07218be3e7825c956e00dfa8c63d08fd2e},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.03509. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Chain-of-thought (CoT) reasoning allows humans to inspect how large language models reach their answers, and oversee model behaviour. This reasoning comes at an increased inference cost, motivating efficient methods that train models to solve tasks using fewer tokens. However, a common concern is that such training may cause models to skip important reasoning steps, so the CoT no longer faithfully reflects the model's decision. It is unclear whether or when this occurs in practice, since different efficiency methods apply length pressure to models' CoT in distinct ways, and faithfully explaining a model's decision takes more tokens on some tasks than others. To understand these dynamics, we fine-tune a variety of models with three methods that apply length pressure differently, namely a fixed generation budget, a per-example length target, and a group-relative length reward. We evaluate how efficient reasoning affects CoT faithfulness (i.e., how well the CoT reflects model decisions on related inputs) and monitorability (i.e., whether the CoT reveals when input interventions alter the output). We find that it affects faithfulness and monitorability differently. Faithfulness falls in most settings, primarily because the trained models are less consistent. Monitorability is more robust, as models keep acknowledging the influence on their answer even when the CoT is much shorter.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2610.03509 — 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 2026-10-05T22:45:55.702Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).