TextSquare: Scaling up Text-Centric Visual Instruction Tuning
This work addresses the lack of extensive, high-quality instruction tuning data for text-centric VQA, which is a bottleneck for open-source MLLMs competing with leading models.
The paper tackles the problem of text-centric visual question answering (VQA) by introducing a massive, high-quality instruction-tuning dataset called Square-10M, generated using closed-source multimodal large language models (MLLMs). The resulting model, TextSquare, achieves 62.2% on OCRBench, outperforms previous open-source state-of-the-art models, and surpasses top-tier models like GPT4V and Gemini in 6 of 10 benchmarks.
Text-centric visual question answering (VQA) has made great strides with the development of Multimodal Large Language Models (MLLMs), yet open-source models still fall short of leading models like GPT4V and Gemini, partly due to a lack of extensive, high-quality instruction tuning data. To this end, we introduce a new approach for creating a massive, high-quality instruction-tuning dataset, Square-10M, which is generated using closed-source MLLMs. The data construction process, termed Square, consists of four steps: Self-Questioning, Answering, Reasoning, and Evaluation. Our experiments with Square-10M led to three key findings: 1) Our model, TextSquare, considerably surpasses open-source previous state-of-the-art Text-centric MLLMs and sets a new standard on OCRBench(62.2%). It even outperforms top-tier models like GPT4V and Gemini in 6 of 10 text-centric benchmarks. 2) Additionally, we demonstrate the critical role of VQA reasoning data in offering comprehensive contextual insights for specific questions. This not only improves accuracy but also significantly mitigates hallucinations. Specifically, TextSquare scores an average of 75.1% across four general VQA and hallucination evaluation datasets, outperforming previous state-of-the-art models. 3) Notably, the phenomenon observed in scaling text-centric VQA datasets reveals a vivid pattern: the exponential increase of instruction tuning data volume is directly proportional to the improvement in model performance, thereby validating the necessity of the dataset scale and the high quality of Square-10M.