LLaVA-NeXT-Interleave: Tackling Multi-image, Video, and 3D in Large Multimodal Models

TL;DR

LLaVA-NeXT-Interleave advances large multimodal models by enabling them to handle multi-image, video, and 3D tasks simultaneously, with a comprehensive dataset and benchmark, demonstrating superior performance and emerging cross-modal capabilities.

Contribution

It introduces a unified framework and dataset for multi-image, video, and 3D tasks in LMMs, enabling cross-scenario generalization and new capabilities.

Findings

Achieves leading results on multi-image, video, and 3D benchmarks.

Maintains performance on single-image tasks.

Exhibits emerging cross-modal transfer capabilities.

Abstract

Visual instruction tuning has made considerable strides in enhancing the capabilities of Large Multimodal Models (LMMs). However, existing open LMMs largely focus on single-image tasks, their applications to multi-image scenarios remains less explored. Additionally, prior LMM research separately tackles different scenarios, leaving it impossible to generalize cross scenarios with new emerging capabilities. To this end, we introduce LLaVA-NeXT-Interleave, which simultaneously tackles Multi-image, Multi-frame (video), Multi-view (3D), and Multi-patch (single-image) scenarios in LMMs. To enable these capabilities, we regard the interleaved data format as a general template and compile the M4-Instruct dataset with 1,177.6k samples, spanning 4 primary domains with 14 tasks and 41 datasets. We also curate the LLaVA-Interleave Bench to comprehensively evaluate the multi-image performance of LMMs. Through extensive experiments, LLaVA-NeXT-Interleave achieves leading results in multi-image, video, and 3D benchmarks, while maintaining the performance of single-image tasks. Besides, our model also exhibits several emerging capabilities, e.g., transferring tasks across different settings and modalities. Code is available at https://github.com/LLaVA-VL/LLaVA-NeXT