arXiv:2608. 14647v1 Announce Type: cross Abstract: Dirty-history rollouts make multi-turn on-policy self-distillation (OPSD) brittle: once a student emits an erroneous intermediate reply, later turns are conditioned on that reply, and uniform distillation can spend loss on tokens that carry little corrective signal.
arXiv:2607. 07050v3 Announce Type: replace-cross Abstract: Top-K teacher logits make on-policy distillation tractable, but probability mass is not the same as decision support.
The paper introduces Teacher-Gated On-Policy Distillation (TGOPD), a method that verifies teacher reliability at the prompt level before applying dense supervision in on-policy distillation. TGOPD uses verifier-scored teacher probes to decide whether to route a prompt to dense OPD or to a verifier-grounded alternative. Experiments on 4B and 35B models across mathematics, code, and instruction tasks show TGOPD outperforms vanilla OPD and improves teacher GPU utilization from 9.8% to 78.9% in a 4B single-domain run.
The paper investigates on‑policy distillation (OPD), showing that teacher supervision during OPD contains significant noise that grows with teacher size, yet the student policy remains largely unaffected by this noise. It finds that OPD’s gains stem mainly from suppressing low‑log‑probability tokens, a process that can be replicated without a teacher. Building on this insight, the authors propose On‑Policy Self‑Adaptation (OPSA), a supervision‑free method that uses entropy‑adaptive negative advantages to improve performance on several benchmarks, outperforming both the base model and OPD.
arXiv:2605.28791v2 Announce Type: replace-cross Abstract: On-policy self-distillation (SD) improves LLM reasoning by using teacher-side privileged information (PI) to turn sparse verifier outcomes in...
arXiv:2608. 09836v1 Announce Type: new Abstract: On-policy distillation (OPD) has emerged as a core component of modern LLM post-training pipelines, yet we reveal a failure mode: degenerate agreement, where students exploit repetitive loops to achieve near-perfect token agreement with the teacher despite globally flawed responses.
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.
arXiv:2609. 40285v1 Announce Type: new Abstract: On-policy distillation (OPD) is a promising approach for training language agents, providing dense teacher supervision on student-generated trajectories.
arXiv:2608. 06296v1 Announce Type: new Abstract: On-policy (Self-)Distillation (OPD / OPSD) has shown strong potential for post-training large language models (LLMs).
arXiv:2606. 26091v1 Announce Type: new Abstract: On-policy self-distillation achieves strong pass@1 accuracy by using a single model as both teacher and student, with the teacher conditioned on a correct demonstration to provide dense token-level feedback.
arXiv:2608. 10905v1 Announce Type: new Abstract: On-policy distillation (OPD) applies token-level teacher supervision to student-generated trajectories, but this supervision is not always reliable.
The paper introduces VISTA, a method that enhances on‑policy self‑distillation (OPSD) by adapting the teacher model toward the student’s distribution using outcome‑verified rollouts. VISTA keeps the standard OPSD student update but selectively adjusts the teacher only on the top‑k positions with the largest teacher‑student KL divergence, without adding new sampling or reward objectives. Experiments on AIME24, AIME25, and HMMT25 with Qwen3 models show that VISTA outperforms OPSD across all scales, improving Avg@12 by up to 2.1 points.
arXiv:2607. 29209v1 Announce Type: cross Abstract: Reinforcement learning with verifiable rewards (RLVR) broadcasts a single response-level reward to every token, while on-policy distillation (OPD) scores each token against a stronger teacher for a dense advantage but caps performance at teacher quality and discourages exploration beyond it.