HoneyBee: Data Recipes for Vision-Language Reasoners
This work addresses the challenge of dataset curation for VLMs, which is crucial for advancing vision-language reasoning, but it is incremental as it builds on existing methods with new data strategies.
The paper tackles the problem of constructing effective training datasets for vision-language models (VLMs) by introducing data curation approaches, and finds that strategies like targeted interventions and scaling data dimensions improve reasoning capabilities, resulting in a new dataset (HoneyBee) that boosts VLM performance, e.g., a 3B-parameter model outperforms SOTA by 7.8% on MathVerse.
Recent advances in vision-language models (VLMs) have made them highly effective at reasoning tasks. However, the principles underlying the construction of performant VL reasoning training datasets remain poorly understood. In this work, we introduce several data curation approaches and study their impacts on VL reasoning capabilities by carefully controlling training and evaluation setups. We analyze the effects of context (image and question pair) sources, implement targeted data interventions, and explore scaling up images, questions, and chain-of-thought (CoT) solutions. Our findings reveal that (a) context source strategies significantly affect VLM performance, (b) interventions such as auxiliary signals from image captions and the inclusion of text-only reasoning yield substantial gains, and (c) scaling all data dimensions (e.g., unique questions per image and unique CoTs per image-question pair) consistently improves reasoning capability. Motivated by these insights, we introduce HoneyBee, a large-scale, high-quality CoT reasoning dataset with 2.5M examples consisting 350K image-question pairs. VLMs trained with HoneyBee outperform state-of-the-art models across model sizes. For instance, a HoneyBee-trained VLM with 3B parameters outperforms the SOTA model and the base model by 7.8% and 24.8%, respectively, on MathVerse. Furthermore, we propose a test-time scaling strategy that reduces decoding cost by 73% without sacrificing accuracy. Overall, this work presents improved strategies for VL reasoning dataset curation research.