Zengzhi Wang

Scientific reasoning is critical for developing AI scientists and supporting human researchers in advancing the frontiers of natural science discovery. However, the open-source community has primarily focused on mathematics and coding while neglecting the scientific domain, largely due to the absence of open, large-scale, high-quality, verifiable scientific reasoning datasets. To bridge this gap, we first present **CuratedReasoning**, an open dataset featuring truthful reference answers extracted from 12k high-quality human-written PDFs, comprising 650k reasoning questions spanning 7 scientific disciplines. We further introduce **MegaScience**, a large-scale mixture of high-quality open-source datasets totaling 1.25 million instances, developed through systematic ablation studies that evaluate various data selection methodologies to identify the optimal subset for each publicly available scientific dataset. Meanwhile, we build a comprehensive evaluation system covering diverse subjects and question types across 14 benchmarks, incorporating comprehensive answer extraction strategies to ensure accurate evaluation metrics. Our experiments demonstrate that our datasets achieve superior performance and training efficiency with more concise response lengths compared to existing open-source scientific datasets. Furthermore, we train Llama3.1-8B, Qwen2.5-7B, and Qwen3 series base models on MegaScience, which outperform the corresponding official instruct models in average performance (e.g., +3.24% for Qwen3-30B-A3B). In addition, **MegaScience exhibits greater effectiveness for stronger models, suggesting scaling benefits for scientific tuning**. We release our data curation pipeline, evaluation system, datasets, and nine trained models to the community to advance scientific reasoning research.