Look Ahead Before You Distill: Future Trajectory Validation of Teacher Guidance for Agentic On-Policy Distillation
This work addresses the problem of teacher guidance effectiveness in on-policy distillation for multi-turn agentic tasks, offering a method that improves performance over existing distillation approaches.
The paper introduces FutureBridge-OPD (FTB), a method that validates teacher guidance by executing a short teacher bridge at high-disagreement states to assess its effect on subsequent student trajectories, improving on-policy distillation for agentic tasks. FTB outperforms vanilla OPD and TCOD by an average of 16.6 and 7.6 points on ALFWorld, WebShop, and ScienceWorld, respectively, under a Qwen3-32B teacher to Qwen3-1.7B student setting.
On-policy distillation (OPD) provides teacher supervision on states visited by the student, reducing the distribution gap between training and inference. However, in multi-turn agentic tasks, student deviations may accumulate over time, gradually moving the trajectory away from states where teacher guidance remains effective. Our quantitative analysis further shows that high-disagreement states offer promising opportunities for teacher guidance, but determining whether such guidance is beneficial requires examining its effect on subsequent student trajectories. We propose FutureBridge-OPD (FTB), which executes a short teacher bridge at a high disagreement state and uses the resulting student continuation to assess whether the bridge increases the density of positive distillation signals relative to the teacher. On ALFWorld, WebShop, and ScienceWorld, under the main Qwen3-32B teacher to Qwen3-1.7B student setting, FTB outperforms vanilla OPD and TCOD by an average of 16.6 and 7.6 points, respectively, and remains effective across student scales and teacher settings. Our code is publicly available at https://github.com/ChenChiShui/FutureBridge-OPD.