CLEAT: Learning Value-Guided Clarification for Interactive Agents
Quick Answer
CLEAT introduces a framework for interactive agents that co-trains a clarification-value controller and task policy, improving decision-making by adapting clarification guidance.
Quick Take
With Qwen3-4B, it achieves top scores across ten evaluation settings, outperforming the strongest baseline by up to 13.2 percentage points in UserGym Travel accuracy.
Key Points
- CLEAT co-trains clarification and task policies from shared interaction outcomes.
- Residual query value measures utility gained from asking clarifying questions.
- Qwen3-4B achieved highest scores across four benchmarks with significant performance gains.
- Co-training improves task policy effectiveness even without the controller during inference.
- Clarification guidance adapts as the agent learns, enhancing decision-making.
Paper Resources
📖 Reader Mode
~2 min readAbstract:LLM agents can execute increasingly complex tasks, but effective assistance also requires aligning their actions with goals and constraints that users may leave unstated. We study this problem as interactive intent alignment through clarification. Yet clarification is useful only when the agent can turn the answer into better decisions. As training changes how the agent handles incomplete information and uses additional answers, the same clarification can become more or less useful. We introduce CLEAT (CLarification through valuE-guided Agent co-Training), a framework that co-trains a clarification-value controller and the task policy from shared interaction outcomes. We formulate residual query value to measure the task utility gained by asking rather than proceeding under the current policy, and estimate it through paired rollouts from the same interaction history. These paired returns train the controller to predict clarification value and the task policy to select useful questions and act on the answers. The controller guides subsequent rollout collection, while outcomes from the updated agent provide new value labels that allow clarification guidance to adapt as the agent learns. With Qwen3-4B, CLEAT achieves the highest mean score in all ten evaluation settings across four benchmarks. Gains over the strongest baseline reach 13.2 and 6.5 percentage points on UserGym Travel and IN3 underspecification judgment accuracy, respectively. Ablations show that co-training improves the task policy even when the controller is removed at inference, while adapting clarification guidance to the evolving policy yields further gains.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.12655 [cs.AI] |
| (or arXiv:2610.12655v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.12655 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Jiatan Huang [view email]
[v1]
Thu, 8 Oct 2026 19:22:21 UTC (572 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.AI
See more →VEPA: Verified Evolving Formal Policy for Lifelong Agent Guardrails
VEPA introduces a framework for verified evolving formal policies for LLM agents, enhancing safety by adapting guardrails based on deployment feedback. It outperforms existing methods across benchmarks like ASSEBench-Safety and ASSEBench-Security, ensuring deterministic policy enforcement even under challenging conditions.