MAmmoTH-VL: Eliciting Multimodal Reasoning with Instruction Tuning at Scale

TL;DR

This paper introduces MAmmoTH-VL, a large-scale multimodal instruction-tuning dataset with rich rationales, significantly enhancing reasoning abilities of multimodal large language models and achieving state-of-the-art results.

Contribution

The authors present a scalable method to create a 12 million instruction-response dataset with detailed rationales, improving reasoning in multimodal models beyond prior datasets.

Findings

Achieved state-of-the-art performance on MathVerse, MMMU-Pro, and MuirBench benchmarks.

Significant improvements in reasoning capabilities with up to 13.3% gains.

Key dataset construction components like rewriting and self-filtering are crucial.

Abstract

Open-source multimodal large language models (MLLMs) have shown significant potential in a broad range of multimodal tasks. However, their reasoning capabilities remain constrained by existing instruction-tuning datasets, which were predominately repurposed from academic datasets such as VQA, AI2D, and ChartQA. These datasets target simplistic tasks, and only provide phrase-level answers without any intermediate rationales. To address these challenges, we introduce a scalable and cost-effective method to construct a large-scale multimodal instruction-tuning dataset with rich intermediate rationales designed to elicit CoT reasoning. Using only open models, we create a dataset containing 12M instruction-response pairs to cover diverse, reasoning-intensive tasks with detailed and faithful rationales. Experiments demonstrate that training MLLMs on this dataset significantly improves reasoning capabilities, achieving state-of-the-art performance on benchmarks such as MathVerse (+8.1%), MMMU-Pro (+7%), and MuirBench (+13.3%). Additionally, the model demonstrates notable improvements of up to 4% on non-reasoning-based benchmarks. Ablation studies further highlight the importance of key components, such as rewriting and self-filtering, in the dataset construction process.