Metraipolis › archive › dc_23773be6925e
Do Uncertainty Signals Help? A Systematic Study of Uncertainty-Aware Decoding with Rollback Mechanisms — arXiv:2608.14653
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.14653
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-18T17:51:40.036Z
- SHA-256 of the archived text
c70791943bc892cc64fbf68e128077eba0dd0e1138bebaf0cd690df0bc5bcca8- Stable URL of this page
https://metraipolis.com/documento/dc_23773be6925e
How to cite
Do Uncertainty Signals Help? A Systematic Study of Uncertainty-Aware Decoding with Rollback Mechanisms — arXiv:2608.14653 — arXiv (en), https://arxiv.org/abs/2608.14653 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_23773be6925e (Metraipolis archive dc_23773be6925e, sha256 c70791943bc892cc64fbf68e128077eba0dd0e1138bebaf0cd690df0bc5bcca8)
BibTeX
@misc{dc23773be6925e,
title = {Do Uncertainty Signals Help? A Systematic Study of Uncertainty-Aware Decoding with Rollback Mechanisms — arXiv:2608.14653},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.14653},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_23773be6925e; sha256 c70791943bc892cc64fbf68e128077eba0dd0e1138bebaf0cd690df0bc5bcca8},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.14653. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Prediction uncertainty is a widely adopted metric for quantifying model confidence, with downstream applications spanning model explanation, data selection, and prediction rollback. Despite its demonstrated utility, the potential of uncertainty quantification to enhance code generation in large language models (LLMs) remains largely underexplored, raising a critical question: to what extent can uncertainty serve as an effective signal for improving LLM-based code generation? To answer this question, we study uncertainty-aware rollback decoding, an inference-time strategy that uses uncertainty signals to identify unreliable generation regions and roll back to earlier valid prefixes without retraining the model. We evaluate this framework on seven code LLMs, five code generation benchmarks, and eight token-level uncertainty signals under a unified decoding setup. Our results show that the complete rollback framework improves over equal-budget restart across the evaluated benchmarks and model settings, with gains of up to 0.26 in pass@1 and 0.35 in AvgTestPassRate on functional code generation benchmarks, and an absolute improvement of up to 6.4\% in Patch-Aligned Safe Rate on Dsec-Python. Among the evaluated signals, information-theoretic measures such as token entropy and negative log-likelihood show the most favorable overall trend, frequently achieving the best or near-best results on standard benchmarks. A component-controlled ablation further shows that feedback-guided rollback provides the main improvement, while uncertainty localization provides an additional gain when checking, budget, rollback, and branch decay are held fixed. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-08-18 in cs.AI). Categorie arXiv: Machine Learning (cs.LG) ; Artificial Intelligence (cs.AI). Pagina del lavoro: https://arxiv.org/abs/2608.14653 — 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 2026-10-18T16:29:13.929Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-08-18?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).