paper-with-me

홈 › Papers

Towards Long-window Anchoring in Vision-Language Model Distillation

2025-12-25 · Haoyi Zhou, Shuo Li, Tianyu Chen, Qi Song, Chonghan Gao, Jianxin Li arxiv

While large vision-language models (VLMs) demonstrate strong long-context understanding, their prevalent small branches fail on linguistics-photography alignment for a limited window size. We discover that knowledge distillation improves students' capability as a complement to Rotary Position Embeddings (RoPE) on window sizes (anchored from large models). Building on this insight, we propose LAid, which directly aims at the transfer of long-range attention mechanisms through two complementary components: (1) a progressive distance-weighted attention matching that dynamically emphasizes longer position differences during training, and (2) a learnable RoPE response gain modulation that selectively amplifies position sensitivity where needed. Extensive experiments across multiple model families demonstrate that LAid-distilled models achieve up to 3.2 times longer effective context windows compared to baseline small models, while maintaining or improving performance on standard VL benchmarks. Spectral analysis also suggests that LAid successfully preserves crucial low-frequency attention components that conventional methods fail to transfer. Our work not only provides practical techniques for building more efficient long-context VLMs but also offers theoretical insights into how positional understanding emerges and transfers during distillation.

📄 PDF Abstract BibTeX arXiv:2512.21576

Code (0)

등록된 구현이 없습니다.

Tasks

Long-Context UnderstandingKnowledge Distillation

Similar Papers 제목 키워드 기반

DAPD: Dual-Anchored Policy Distillation

2026-08-03 · Jianyu Wu, Yizhou Wang, Encheng Su, Chen Tang 외 hf

On-policy (self) distillation (OPSD) is increasingly adopted for language-model post-training. It strengthens the teacher with privileged information but can induce a privilege illusion: the student learns privilege-depe…

A Recipe for Long-Context Reasoning in Large Language Models via On-Policy Optimization and Distillation

2026-05-12 · Miguel Moura Ramos, Duarte M. Alves, André F. T. Martins arxiv

Existing approaches to post-train models for long-context tasks face complementary limitations: (i) supervised fine-tuning (SFT) provides stable supervision but suffers from exposure bias; (ii) reinforcement learning met…

Reinforcement Learning

On-Policy Distillation for Vision-Language Model Adaptation, an Effective Paradigm on Low-Quality Multimodal Data

2026-09-09 · Hongyuan Zhang, Xianda Guo, Yanlun Peng, Qianlong Yang 외 arxiv

Knowledge distillation offers an efficient route to transfer a task-adapted vision-language teacher to a compact student. The training target in current vision-language distillation methods is typically constructed from …

Knowledge Distillation

EvoDriveVLA: Evolving Driving VLA Models via Collaborative Perception-Planning Distillation

2026-03-10 · Jiajun Cao, Xiaoan Zhang, Xiaobao Wei, Liyuqiu Huang 외 arxiv

Vision-Language-Action models have shown great promise for autonomous driving, yet they suffer from degraded perception after unfreezing the visual encoder and struggle with accumulated instability in long-term planning.…

Autonomous Driving

EndPrompt: Efficient Long-Context Extension via Terminal Anchoring

2026-05-14 · Han Tian, Luxuan Chen, Xinran Chen, Rui Kong 외 arxiv

Extending the context window of large language models typically requires training on sequences at the target length, incurring quadratic memory and computational costs that make long-context adaptation expensive and diff…