Contrastive On-Policy Distillation (COPD) is a framework that improves on-policy distillation by using a frozen teacher to evaluate student states under two contrasting prompts—one encouraging low reasoning effort and one encouraging high effort. The difference in log‑probabilities between these prompts provides a token‑level advantage signal that guides the student toward more concise and efficient reasoning strategies. Experiments on nine multimodal benchmarks show that COPD reduces reasoning length while maintaining task performance, and the contrastive approach can also be applied to on‑policy self‑distillation, allowing a model to compress its own reasoning without an external teacher.
By Jiacheng Ruan, Jun Tang, Wenzhen Yuan, Ting Liu, Shuai Bai, Dayiheng Liu, Zhibo Yang, Yuzhuo Fu
arXiv:2606. 08432v1 Announce Type: new Abstract: On-policy distillation (OPD) has become a central post-training tool for large language models (LLMs), providing dense per-token teacher supervision along the student's own rollouts.
By Li Jiang, Haoran Xu, Yichuan Ding, Amy Zhang
On-policy Distillation (OPD) supervises a student model on trajectories sampled from its own policy by minimizing the divergence between the output distributions of the teacher and student at each token position, thereby providing dense token-level supervision. Although existing OPD methods have demonstrated strong performance in improving the reasoning ability of student models, their objectives fundamentally rely on token-level distribution matching.
Teacher Should Think Ahead: Adaptive Continuations for Reliable On-Policy Distillation explores how on‑policy distillation (OPD) can be improved by addressing teacher uncertainty that contracts as the teacher continues from a student‑generated prefix. The authors identify Teacher Uncertainty Contraction (TUC) and theoretically analyze its variance‑bias trade‑off, leading to the proposal of Adaptive‑Continuations On‑Policy Distillation (AC‑OPD). Experiments on mathematical reasoning and code generation show that AC‑OPD consistently outperforms standard OPD, with controlled‑continuation and matched‑budget analyses supporting the adaptive‑continuation design.
By Jingang Zhou, Yuyi Zhou, Haiyang Guo, Xukai Wang, Shuai Feng, Sirui Gao, Jian Xu, Qingpei Guo, Xu-Yao Zhang
arXiv:2609.33455v2 Announce Type: replace
Abstract: On-policy distillation (OPD) trains a student model on its own trajectories using dense token-level feedback from a stronger teacher model. Since e...
By Zizhuo Lin, Quanling Liu, Yi Yang, Yawei Luo
The paper introduces Student‑Prefix Teacher Adaptation (SPTA), a method that trains a teacher model to better handle student‑generated prefixes before performing on‑policy distillation (OPD). By collecting prefixes of random lengths and using reinforcement learning with verifiable rewards (RLVR), the teacher learns to recover from student‑induced states rather than simply imitate the student’s reasoning. Experiments on mathematical reasoning benchmarks show that SPTA improves student performance compared to standard OPD and OPD with task‑level RLVR‑adapted teachers.
By Xue Zhang, Songming Zhang, Yunlong Liang, Fandong Meng, Yufeng Chen, Jinan Xu
arXiv:2609.27421v1 Announce Type: new
Abstract: Multi-constraint instruction following requires a model to respond to a query under many simultaneously active constraints. Even strong instruction-tun...
By Yanzhao Zheng, Yuanqiang Yu, Tianze Xu, Chao Ma, Zhentao Zhang, Jihuai Zhu, Baohua Dong, Hangcheng Zhu, Ruohui Huang
The paper proposes Mixture Proposal On-Policy Distillation (MPOPD) to address a mismatch issue in on‑policy distillation (OPD) where the rollout policy is fixed to the current student. By treating the rollout policy as an optimization target, MPOPD derives a geometric mixture of student and teacher policies and introduces a proposal policy for efficient rollout generation. Experiments on 16 reasoning benchmarks show consistent gains over OPD, notably a 6.91% accuracy improvement on math reasoning with Qwen3-4B-Base and a 41% reduction in generation length.
By Sirui Chen, Yunzhe Qi, Haobo Xu, Jiaru Zou, Mengting Ai, Zihao Li, Jingrui He
arXiv:2609.38025v1 Announce Type: cross
Abstract: On-policy distillation (OPD) trains a student on its own generated responses using dense, token-level supervision from a stronger teacher. Vanilla OP...
By Zhenyu Wang, Tianze Wang, Linjun Zhang, Yifan Hu
On-policy distillation (OPD) trains a student on its own generated responses using dense, token-level supervision from a stronger teacher. Vanilla OPD treats all teacher signals equally, assuming that...
The paper investigates on‑policy distillation (OPD), a method that trains a student model using its own rollouts while a teacher remains frozen, to see if it can help a student improve on problems it currently fails. Experiments with Qwen‑3 and Phi‑4 models on a competition‑mathematics benchmark show that OPD actually degrades performance when training focuses on failure cases, even though training loss (KL divergence) continues to drop. However, when OPD is applied to a broader set of problems, it improves accuracy by 4.6 percentage points, indicating that OPD works well when most training data contains correct student prefixes but can reinforce errors if the training regime concentrates on failures.
By Shuo Yang, Zheyu Zhang, Mahito Sugiyama, Gjergji Kasneci
The paper investigates the learning dynamics of on‑policy distillation (OPD), a method that combines teacher supervision with training on student‑generated trajectories. It identifies two stages of the feedback loop: converting the distillation signal into parameter updates, and broadcasting those updates to alter predictions and reshape the rollout distribution. The study explains why students may be unresponsive to teacher disagreement and how updates that improve local agreement can lead to degeneration by increasing the likelihood of poor continuations. It also shows that offline distillation can prepare a student for OPD without boosting immediate accuracy, yet OPD can still yield significant accuracy gains after sufficient offline preparation.
By Guan Zhe Hong, Sumeet Ramesh Motwani, Liang Chen, Yangchen Pan, Junchi Yu, Jialin Yu, Philip Torr, Yi Ren