English

QLIP: A Dynamic Quadtree Vision Prior Enhances MLLM Performance Without Retraining

Machine Learning 2026-03-27 v2

Abstract

Multimodal Large Language Models (MLLMs) encode images into visual tokens, aligning visual and textual signals within a shared latent space to facilitate crossmodal representation learning. The CLIP model is a widely adopted foundational vision language model whose vision encoder has played a critical role in the development of MLLMs such as LLaVA. However, the CLIP vision encoder suffers from notable limitations including being constrained to only handling fixed input resolutions and a failure to produce separated embeddings for dissimilar images. Replacing the vision encoder of an existing model typically incurs substantial computational costs because such a change often necessitates retraining the entire model pipeline. In this work, we identify two factors which underlie the limitations of the CLIP vision encoder: mesoscopic bias and interpolation bias. To address these issues, we propose QLIP, a drop-in replacement for CLIP that can be seamlessly integrated with existing MLLMs with only a few lines of code and can enhance both coarse-grained and fine-grained visual understanding, without re-training. QLIP is designed around an image quadtree which replaces the standard uniform grid patches with a novel content aware patchification. Our experimental results demonstrate that QLIP improves the general visual question answering accuracy of the LLaVA v1.5 model series across various model sizes--without requiring retraining or fine-tuning of the full MLLM. Notably, QLIP boosts detailed understanding performance on the challenging V-star benchmark by up to 13.6 percent.

Keywords

Cite

@article{arxiv.2505.23004,
  title  = {QLIP: A Dynamic Quadtree Vision Prior Enhances MLLM Performance Without Retraining},
  author = {Kyle R. Chickering and Bangzheng Li and Muhao Chen},
  journal= {arXiv preprint arXiv:2505.23004},
  year   = {2026}
}

Comments

Accepted as ICLR 2026 poster. 22 pages, 19 figures

R2 v1 2026-07-01T02:47:38.074Z