Prior-as-Skill: Remember Outside and Read Inside with On-Policy Self-Distillation for Self-Evolving VLM Agents
cs.CV, cs.AI
Submitted: 2026-11-11
Updated: 2026-11-11
Terminology
Sources
- Qwen3-VL Technical Report
- Qwen2.5-VL Technical Report
- Thinking Without Images: Internalizing Visual Manipulation with On-Policy Self-Distillation
- Rethinking Continual Experience Internalization for Self-Evolving LLM Agents
- Mitigating the Influence of Distractor Tasks in LMs with Prior-Aware Decoding
- Distilling the Knowledge in a Neural Network
- Building and better understanding vision-language models: insights and future directions
- Learning by Distilling Context
- GATES: Self-Distillation under Privileged Context with Consensus Gating
- Distill the Visual Evidence, Not Just the Answer: Cross-World On-Policy Distillation for Vision-Language Models
- UCOB: Learning to Utilize and Evolve Agentic Skills via Credit-Aware On-Policy Bidirectional Self-Distillation
- AtlasVA: Self-Evolving Visual Skill Memory for Teacher-Free VLM Agents
- InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiency
- When Context Returns: Toward Robust Internalization in On-Policy Distillation
- SEED: Self-Evolving On-Policy Distillation for Agentic Reinforcement Learning
- OPID: On-Policy Skill Distillation for Agentic Reinforcement Learning
- On-Policy Context Distillation for Language Models
- LatentSkill: From In-Context Textual Skills to In-Weight Latent Skills for LLM Agents
Related papers
- TEA: Text Encoder Alignment for Robust Concept Erasure in Text-to-Image Models
- Do Egocentric Video-Language Models Capture Both Hand- and Object-Centric Cues?
- Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment
- PVRF: All-in-one Adverse Weather Removal via Prior-modulated and Velocity-constrained Rectified Flow
- DLWM: Diverse Latent World Models for Efficient Multimodal Reasoning
- Out of Sight, Still in Mind: Token Compression for Omni-LLMs