Multimodal classifiers can converge to modality-dominant solutions in which one modality dominates the joint prediction, suppressing the learning of others. Existing balancing methods mainly adjust losses, gradients, or modality contributions, largely treating modality imbalance as an optimization problem while implici...
Ze-Chang Xiong, Da Li, Rong Yin et al.· 0 citations
This paper introduces AutoSkill, a source-supervised framework for automatically discovering and routing executable frame-selection skills, and shows that the relative effectiveness of frame-selection strategies varies across semantic categories and benchmarks, motivating adaptive evidence acquisition.
This work proposes SHIFT, a retrieval training framework based on LLMs that transfers LLMs into reasoning-efficient retrievers with residual projection and task-oriented bidirectional attention aggregation in the latent space, and alleviates the mismatch between contrastive learning and implicit reasoning using fine-gr...
Yuxiao Luo, Da Li, Mingjie Zhang et al.· arXiv.org· 0 citations
This work constructed Illusion-Reasoning, a benchmark of illusion images collected from the real world, incorporating diverse annotated question-answer pairs, and shows that the reasoning capabilities of a wide range of LVLMs are not as advanced as claimed.
