Metraipolis › archive › dc_dcbed2976391
Pareto-Dominant Clarification: Post-Training Coding LLMs via PPO-Lagrangian Budget Constraints — arXiv:2610.04089
- Source
- arXiv
- Original
- https://arxiv.org/abs/2610.04089
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-06T05:30:24.211Z
- SHA-256 of the archived text
4f1081a825f115ada9b6d033e4323da7af7deb8ac175ee916d92475d42785fee- Stable URL of this page
https://metraipolis.com/documento/dc_dcbed2976391
How to cite
Pareto-Dominant Clarification: Post-Training Coding LLMs via PPO-Lagrangian Budget Constraints — arXiv:2610.04089 — arXiv, https://arxiv.org/abs/2610.04089 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_dcbed2976391 (Metraipolis archive dc_dcbed2976391, sha256 4f1081a825f115ada9b6d033e4323da7af7deb8ac175ee916d92475d42785fee)
BibTeX
@misc{dcdcbed2976391,
title = {Pareto-Dominant Clarification: Post-Training Coding LLMs via PPO-Lagrangian Budget Constraints — arXiv:2610.04089},
howpublished = {arXiv},
url = {https://arxiv.org/abs/2610.04089},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_dcbed2976391; sha256 4f1081a825f115ada9b6d033e4323da7af7deb8ac175ee916d92475d42785fee},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv, https://arxiv.org/abs/2610.04089. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Coding agents operating under ambiguous instructions or user prompts must decide whether to ask clarifying questions or attempt a solution directly. While clarification from the user may improve the correctness of the agent's solution, each back-and-forth interaction incurs user and system costs, forming an explicit accuracy vs. efficiency tradeoff. Existing works study clarification behavior but do not train policies under enforceable clarification budgets; penalty-based approaches typically require separate coefficient tuning swept across all clarification budget levels. We formulate clarification as a Constrained Markov Decision Process (CMDP) and post-train Qwen2.5-Coder-7B-Instruct with PPO-Lagrangian to optimize coding accuracy, subject to an expected question-budget constraint. Evaluated on HumanEvalComm with a GPT-4o-mini oracle simulator, the resulting policies reveal that untuned clarification behavior is Pareto-inefficient: budget-constrained policies can simultaneously achieve higher accuracy and lower clarification rates than the baseline model. Across budget levels, we observe a log-shaped Pareto frontier with diminishing returns to additional clarification. Gains arise not from simply asking more questions overall, but from improved question targeting and better code generation under ambiguity. Without explicit supervision, trained policies learn to allocate clarification budget non-uniformly, asking more frequently on tougher (multi-degradation) tasks. These results suggest that unconstrained interactive LLM systems may systematically use clarification inefficiently. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2610.04089 — 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 2026-10-06T05:27:05.653Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).