MatCha: Enhancing Visual Language Pretraining with Math Reasoning and Chart Derendering
This addresses a specific bottleneck in visual language modeling for tasks involving charts and plots, showing incremental improvements with broader applicability.
The paper tackles the problem of poor performance by state-of-the-art vision-language models on visual language data like charts and plots, proposing MatCha pretraining to enhance capabilities in chart derendering and math reasoning, resulting in up to nearly 20% improvement on benchmarks like PlotQA and ChartQA.
Visual language data such as plots, charts, and infographics are ubiquitous in the human world. However, state-of-the-art vision-language models do not perform well on these data. We propose MatCha (Math reasoning and Chart derendering pretraining) to enhance visual language models' capabilities in jointly modeling charts/plots and language data. Specifically, we propose several pretraining tasks that cover plot deconstruction and numerical reasoning which are the key capabilities in visual language modeling. We perform the MatCha pretraining starting from Pix2Struct, a recently proposed image-to-text visual language model. On standard benchmarks such as PlotQA and ChartQA, the MatCha model outperforms state-of-the-art methods by as much as nearly 20%. We also examine how well MatCha pretraining transfers to domains such as screenshots, textbook diagrams, and document figures and observe overall improvement, verifying the usefulness of MatCha pretraining on broader visual language tasks.