cs.CVNov 11, 2026

Prior-as-Skill: Remember Outside and Read Inside with On-Policy Self-Distillation for Self-Evolving VLM Agents

Authors: Yan Lin, Jingyu Sun, Yifei Qian, Zhengtao Yao, Yuxia Wang, Daizong Liu, Chun Pong Lau, Zhongliang Guo, 

Organizations: Newcastle University · University of Manchester · University of Melbourne · University of Nottingham · 

Abstract

Self-evolving visual agents keep what they learn as skills: key frames with step notes on a task's steps and their order. On-policy self-distillation (OPSD) is now a popular way to train such skills into a vision-language model that matches, on its own answers, a teacher that also reads the written order. At deployment, frames often go missing or show an unusual order, as when a person or a website reorders the steps. Existing internalization methods keep any order they store in the frame-reading weights: trained on full frames, the model guesses when they are missing, and trained with frames withheld, it stores the order, so memory can override what it sees. We propose PriorSplit, which fills in missing frames from an order kept outside the model, with no training beyond OPSD. It counts which of two neighboring steps comes first in training episodes, and the model checks if both asked steps are in view: if so, it answers from the frames, and otherwise the counts answer. Extensive experiments on synthetic skills with five models and on step-tagged order questions in new episodes of seen web, cooking and egocentric tasks demonstrate that, with frames missing, PriorSplit outperforms every compared method at OPSD's training cost. With Qwen3-VL-4B on synthetic skills, it also stays close to OPSD on unusual orders, where methods that gain most by storing the order fall behind.

Figures & tables

Appendix figures & tables20 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. Do As I Do, Not As I Say: On-Policy Distillation Inherits the Teacher in Vision-Language Models

    Oct 14, 2026Frank F. Yang, Kefei Yan, Yiren Wang +1VLM DistillationOn-Policy Distillation

  2. Self-Boosting Vision-Language Models with Noisy Student On-Policy Self-Distillation

    Jul 25, 2026Shuai Wang, Daoan Zhang, Zhe Tang +2VLM AdaptationOn-Policy Self-Distillation

  3. Where-OPD: Spatially Guided On-Policy Self-Distillation of MLLMs with Synthetic Scenes

    Oct 1, 2026Sophia Sirko-Galouchenko, Monika Wysoczanska, Andrei Bursuc +2VLM DistillationMultimodal Large Language Models