Perceiver-VL: Efficient Vision-and-Language Modeling with Iterative Latent Attention

TL;DR

Perceiver-VL introduces an efficient vision-and-language framework utilizing iterative latent cross-attention, achieving scalable multimodal processing with reduced computational complexity and competitive performance on various benchmarks.

Contribution

It presents a novel scalable multimodal model with linear complexity using iterative latent attention and explores efficiency improvements like LayerDrop and mixed-stream architecture.

Findings

Achieves lowest GFLOPs and latency among benchmarks.

Maintains competitive performance with high efficiency.

Provides extensive analysis of model components and strategies.

Abstract

We present Perceiver-VL, a vision-and-language framework that efficiently handles high-dimensional multimodal inputs such as long videos and text. Powered by the iterative latent cross-attention of Perceiver, our framework scales with linear complexity, in contrast to the quadratic complexity of self-attention used in many state-of-the-art transformer-based models. To further improve the efficiency of our framework, we also study applying LayerDrop on cross-attention layers and introduce a mixed-stream architecture for cross-modal retrieval. We evaluate Perceiver-VL on diverse video-text and image-text benchmarks, where Perceiver-VL achieves the lowest GFLOPs and latency while maintaining competitive performance. In addition, we also provide comprehensive analyses of various aspects of our framework, including pretraining data, scalability of latent size and input size, dropping cross-attention layers at inference to reduce latency, modality aggregation strategy, positional encoding, and weight initialization strategy. Our code and checkpoints are available at: https://github.com/zinengtang/Perceiver_VL