When Should the Teacher Move? Temporal Coupling and Stability in Self On-Policy Distillation
About
Self on-policy distillation trains a student policy against a teacher derived from its own parameter history, yet the teacher's update schedule -- which governs the \emph{temporal coupling} between teacher and student -- has not been systematically studied as a stability variable. Through a controlled schedule sweep on Qwen3-8B, we establish that \emph{isolation periods}, defined as complete teacher freezing between updates, are the key structural property enabling stable learning, not teacher age. To characterize these underlying training dynamics, we introduce a diagnostic framework of temporal KL structure, refresh shock, and length-tail risk. This framework further uncovers \emph{state-oblivious collapse}: optimal short-horizon fixed schedules catastrophically fail under long-horizon training because a clock-driven refresh can copy a transiently drifting student into the teacher in a single, irreversible step. This failure mode is invisible under short-horizon evaluation and mechanistically distinct from EMA's chronic contamination. To address this, we propose \emph{Consolidation-Gated Teacher Refresh} (CGTR), which preserves isolation periods while gating each refresh on joint evidence of reward improvement and length-tail safety, ensuring every teacher movement responds to genuine student consolidation rather than a clock signal. With a single shared parameter set and no per-dataset retuning, CGTR achieves \textbf{zero collapse} and the best final score on all four tasks (Chemistry, Biology, Physics, ToolUse), self-regulating its refresh frequency to each task's learning dynamics.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Cross-task generalization | Chemistry | Test Final Score0.806 | 3 | |
| Cross-task generalization | Biology | Test Fin Score48.5 | 3 | |
| Cross-task generalization | Physics | Test Final Score70.3 | 3 | |
| Cross-task generalization | ToolUse | Test Final Score53.6 | 3 | |
| Multiple-Choice Reasoning | SciKnowEval Chemistry L3 (test) | Final Test Mean Score@1680.6 | 3 |