OPD‑Aha is a privileged on‑policy distillation method that improves multimodal reasoning by reconstructing the distillation target from the teacher’s isolated visual preference instead of relying on fragile teacher‑student discrepancies. It suppresses continuations that contradict the image, encouraging students to interrupt flawed reasoning with reflection tokens such as "wait" and "actually." This approach leads to consistent improvements across fine‑grained perception and complex multimodal reasoning benchmarks.
By Chenhao Qiu, Dawei Li, Yechao Zhang, Lei Gong, Zhen Tan
arXiv:2609.36838v1 Announce Type: cross
Abstract: Visual agents solve problems by interleaving reasoning with image operations, and on-policy distillation (OPD) provides guidance from a strong teache...
By Shaohang Wei, Feifan Song, Guangyue Peng, Wenhao Yu, Wei Li, Wen Luo, Yang Xu, Yufan Shen, Luke Mao, Yang Du, Asher Qin, Houfeng Wang
arXiv:2606. 05718v1 Announce Type: cross Abstract: On-policy distillation (OPD) improves reasoning by training a student on trajectories sampled from its own policy under supervision from a teacher.
By Kanghui Tian, Siyuan Liu, Ziang Yan, Sheng Xia, Shuai Dong, Yi Wang
arXiv:2610.14939v1 Announce Type: new
Abstract: GRPO-style reinforcement learning has advanced vision-language reasoning, but its sequence-level rewards provide limited credit assignment for individu...
By Jiahe Feng
arXiv:2609.39120v1 Announce Type: new
Abstract: On-policy distillation (OPD) improves reasoning by providing token-level supervision from a teacher on a student's own trajectories. Existing methods p...
By Siyuan Liu, Kanghui Tian, Yue Duan, Yutao He, Shangdong Yang, Jian Zhang, Yinghuan Shi
LEGO-OPD introduces a factorized teacher composition for multimodal on‑policy distillation, combining a Language Expert and a Grounding Expert into a single teacher distribution. By treating the language expert as a prior over tokens and the grounding expert as a visual likelihood that updates this prior, the method decouples language reasoning from visual grounding. Adaptive calibration further adjusts the influence of visual evidence at each decoding prefix, preventing over‑ or under‑supervision. Experiments with Qwen3 models demonstrate that LEGO‑OPD outperforms both single‑ and multi‑teacher baselines on multimodal and text‑only reasoning tasks, improving visual perception while preserving language reasoning.
By Jaeyun Shin, Hangeol Chang, Jong Chul Ye
Multimodal on-policy distillation (OPD) aims to improve visual grounding while preserving the strong reasoning capabilities of language models. Recent multi-teacher approaches combine LLM and VLM teac...
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:2608. 05131v1 Announce Type: cross Abstract: On-Policy Self-Distillation (OPSD) has become a standard post-training approach for improving visual reasoning in multimodal large language models (MLLMs).
By Aniri, Jinhe Bi, Peng Liao, Zengjie Jin, Volker Tresp, Fei Shen, Yunpu Ma, Tat-Seng Chua
Unified multimodal models (UMMs) interleave generated ''visual thoughts'' (VTs) with text reasoning to improve spatial tasks. This incurs roughly an order-of-magnitude inference cost from multi-step diffusion.
The paper introduces Contrastive Vision On‑policy Distillation (CV‑OPD), a method that enhances on‑policy distillation for vision‑language models by adding a contrastive divergence loss and a discrepancy‑aware gating mechanism. CV‑OPD contrasts teacher distributions from positive and negative visual crops, forcing the student to rely on genuine visual evidence rather than shortcuts. Experiments on fine‑grained visual benchmarks show that CV‑OPD reduces shortcut learning and improves visual reasoning across existing VLMs.
By Haowei Wang, Hao Lin, Qi Zhu, Jiayi Ji, Deyi Ji, Liujuan Cao, Rongrong Ji
arXiv:2606.18974v3 Announce Type: replace
Abstract: Unified multimodal models (UMMs) interleave generated ''visual thoughts'' (VTs) with text reasoning to improve spatial tasks. This incurs roughly a...
By Pengyu Li, Zhitao Gao, Lingling Zhang, Muye Huang, Yuanming Li, Fangzhi Xu, Jun Liu