arXiv:2606. 07082v1 Announce Type: cross Abstract: On-policy distillation (OPD) is increasingly used to improve large language model reasoning, but its training dynamics remain poorly understood.
arXiv:2606. 13657v2 Announce Type: replace Abstract: On-policy distillation (\textsc{OPD}) has recently become a prominent post-training recipe by combining two desirable ingredients: on-policy student trajectories and dense teacher supervision.
arXiv:2609.36262v1 Announce Type: new Abstract: Recent studies have observed that parameter changes during language-model post-training can be concentrated in a small subset of coordinates. This phen...
arXiv:2605. 25582v2 Announce Type: replace Abstract: Reinforcement learning for large language models faces a fundamental trade-off between sample efficiency and asymptotic performance: strictly on-policy methods discard trajectories after a single update, while off-policy reuse introduces distribution mismatch that existing trust-region techniques mitigate primarily by enforcing conservative optimization, often leaving rich training signals underutilized.
arXiv:2608. 03632v1 Announce Type: new Abstract: On-Policy distillation (OPD) transfers teacher capabilities by supervising student-sampled trajectories with dense token-level teacher signals.
arXiv:2608.24696v1 Announce Type: cross Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) and on-policy distillation (OPD) have become two widely adopted paradigms for post-training lar...
arXiv:2607. 05394v1 Announce Type: cross Abstract: Reinforcement learning with verifiable rewards (RLVR) is a powerful recipe for improving language-model reasoning, but it is expensive to repeat on every new strong model because the target model must generate many rollouts during training.
arXiv:2609.08341v1 Announce Type: new Abstract: On-Policy Distillation (OPD) facilitates the transfer of knowledge from domain expert to student in the post-training phase of Large Language Models (L...
Reinforcement learning with verifiable rewards (RLVR) is a powerful recipe for improving language-model reasoning, but it is expensive to repeat on every new strong model because the target model must generate many rollouts during training. As models scale, post-training itself becomes a bottleneck.
arXiv:2606. 31813v1 Announce Type: cross Abstract: Low-rank adaptation (LoRA) and its variants enable parameter-efficient fine-tuning of large language models under the supervised fine-tuning (SFT) paradigm.
The paper introduces Selective Supervision for Direct-OPD (S$^2$D-OPD), a refinement of Direct On-Policy Distillation that filters out states where the teacher’s policy change is minimal, as measured by the teacher‑reference Jensen‑Shannon divergence. By masking low‑divergence states and keeping only the top 10% of states per response, S$^2$D-OPD improves held‑out accuracy on AIME and HMMT benchmarks across multiple teacher‑student pairs without additional forward passes.
PMOPD introduces a projection-based approach to multi-teacher on-policy distillation, addressing the capability seesaw problem by constructing subspace memories from task-specific parameter displacements and projecting gradients and optimizer updates to avoid cross-task interference. It also includes a lightweight conflict probe for task interaction analysis, a task ordering strategy, and a cycling mechanism to balance subspace estimation and task revisitation. Experiments on Code, Reason, and Math tasks demonstrate that PMOPD improves all evaluated capabilities, raising average scores by 2.54 points on Qwen2.5-7B and 2.09 points on Llama-3.1-8B.
RISE (Recursive Improvement via Self-Extrapolating Policy Distillation) is a new method that builds a synthetic teacher from a language model’s own RLVR training trajectory. By extrapolating the displacement between the current checkpoint and a trailing anchor in parameter or logit space, RISE transforms sparse outcome-based updates into dense token-level targets without external models or privileged conditioning. The approach recursively refines the student model, combining RLVR and on‑policy distillation, and demonstrates superior performance across mathematical reasoning, STEM, code generation, and multi‑turn agentic tasks.