TISD: On-Policy Self-Distillation with Trajectory Intervention

The paper introduces TISD, a trajectory-intervention self-distillation method that forces a teacher-selected branch action and then lets the student generate the suffix, distilling the full trajectory under a privileged-context-conditioned teacher. This approach addresses a data-collection bottleneck in on‑policy self‑distillation by exposing successor contexts that the student would otherwise miss. Experiments on coding and science domains show modest but consistent improvements in average performance metrics compared to baseline methods.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

• efficiency