CoVe: Training Interactive Tool-Use Agents via Constraint-Guided Verification

TL;DR

Results indicate that CoVe provides an effective and efficient pathway for synthesizing training data for state-of-the-art interactive tool-use agents and is competitive with models up to $17\times$ its size.

Abstract

Developing multi-turn interactive tool-use agents is challenging because real-world user needs are often complex and ambiguous, yet agents must execute deterministic actions to satisfy them. To address this gap, we introduce \textbf{CoVe} (\textbf{Co}nstraint-\textbf{Ve}rification), a post-training data synthesis framework designed for training interactive tool-use agents while ensuring both data complexity and correctness. CoVe begins by defining explicit task constraints, which serve a dual role: they guide the generation of complex trajectories and act as deterministic verifiers for assessing trajectory quality. This enables the creation of high-quality training trajectories for supervised fine-tuning (SFT) and the derivation of accurate reward signals for reinforcement learning (RL). Our evaluation on the challenging $τ^2$-bench benchmark demonstrates the effectiveness of the framework. Notably, our compact \textbf{CoVe-4B} model achieves success rates of 43.0\% and 59.4\% in the Airline and Retail domains, respectively; its overall performance significantly outperforms strong baselines of similar scale and remains competitive with models up to $17\times$ its size. These results indicate that CoVe provides an effective and efficient pathway for synthesizing training data for state-of-the-art interactive tool-use agents. To support future research, we open-source our code, trained model, and the full set of 12K high-quality trajectories used for training.

Figures (4)

• Figure 1: Performance evaluation on the $\tau^2$-bench Airline and Retail domains. We report success rates across $pass^1$ to $pass^4$ metrics to assess both peak performance and stability.
• Figure 2: The CoVe framework. Explicit constraints are sampled and fuzzified to guide a User Simulator LLM in generating ambiguous, realistic queries. Upon conversation completion, the original deterministic constraints act as a checklist to automatically verify the agent's tool invocations.
• Figure 2: Performance comparison ($pass^1$) for the Qwen3-4B-Instruct-2507 model fine-tuned on different SFT datasets. Bold and underline denote the best and second-best results.
• Figure 3: The system prompt (top) and the first user prompt (bottom) for the User Simulator. After receiving the first user prompt, the simulator begins issuing task requests to the Agent. In subsequent dialogue turns, the Agent's response serves as the user message for the User Simulator.