Search papers, labs, and topics across Lattice.
The paper introduces MODIX, a training-free framework that dynamically adjusts positional strides in VLMs based on the information density of each modality. MODIX uses covariance-based entropy for intra-modal density and cross-modal alignment for inter-modal interaction to derive scores that rescale positional indices. Experiments show MODIX improves multimodal reasoning and adaptively reallocates attention without modifying model parameters or architecture.
VLMs can achieve better multimodal reasoning simply by dynamically rescaling positional indices based on information density, without any training or architectural changes.
Vision-Language Models (VLMs) have achieved remarkable progress in multimodal understanding, yet their positional encoding mechanisms remain suboptimal. Existing approaches uniformly assign positional indices to all tokens, overlooking variations in information density within and across modalities, which leads to inefficient attention allocation where redundant visual regions dominate while informative content is underrepresented. We identify positional granularity as an implicit resource and propose MODIX (Multimodal Information-Driven Positional IndeX Scaling), a training-free framework that dynamically adapts positional strides based on modality-specific contributions. MODIX jointly models intra-modal density via covariance-based entropy and inter-modal interaction via cross-modal alignment to derive unified scores, which rescale positional indices to allocate finer granularity to informative modalities while compressing redundant ones, without requiring any modification to model parameters or architecture. Experiments across diverse architectures and benchmarks demonstrate that MODIX consistently improves multimodal reasoning and adaptively reallocates attention according to task-dependent information distributions, suggesting that positional encoding should be treated as an adaptive resource in Transformers for multimodal sequence modeling.
