Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

CoLVR: Enhancing Exploratory Latent Visual Reasoning via Contrastive Optimization

About

Due to the potential for exploratory reasoning of Latent Visual Reasoning, recent works tend to enable MLLMs (Multimodal Large Language Models) to perform visual reasoning by propagating continuous hidden states instead of decoding intermediate steps into discrete tokens. However, existing works typically rely on hard alignment objectives to force latent representations to match predefined visual features, thereby severely limiting the exploratory of latent reasoning process. To address this problem, we propose CoLVR (Contrastive Optimization for Latent Visual Reasoning). To obtain a more exploratory visual reasoning, CoLVR introduces a latent contrastive training framework. Firstly, CoLVR learns diverse and exploratory representations with a latent contrastive objective guided by angle-based perturbation, which expands the semantic latent space and avoids over-constrained embedding. Then, CoLVR employs a latent trajectory contrastive reward for RL (Reinforcement Learning) post-training to enable fine-grained optimization of latent visual reasoning process and thus fostering diverse reasoning behaviors. Experiments demonstrate that CoLVR significantly enhances the exploratory capability of latent representations, achieving average improvements of 5.83% on VSP and 8.00% on Jigsaw, while also outperforming existing latent models on out of domain benchmarks, with a 3.40% gain on MMStar. The data, codes, and models are released at https://github.com/Oscar-dzy/CoLVR.

Ziyang Ding, Linjian Meng, Yiming Wu, Yuhan Li, Yuhao Liu, Zhen Zhao• 2026

Related benchmarks

TaskDatasetResultRank
Visual ReasoningJigsaw
Accuracy78
40
Vision UnderstandingMMVP
Accuracy72
36
Visual UnderstandingMMStar--
16
Visual UnderstandingCV-Bench
Accuracy76.95
15
Visual UnderstandingVisPuzzle
Accuracy37
14
Visual ReasoningTertis
Accuracy44.67
9
Visual Spatial PerceptionVSP Unseen
Accuracy (Level 7)43
9
Visual Spatial PerceptionVSP Total
Accuracy (Total)65.83
9
Visual Spatial PerceptionVSP Seen
Accuracy (Level 3)94
9
Visual UnderstandingV*
Accuracy80.63
3
Showing 10 of 10 rows

Other info

Follow for update