News
Neutral
Tether Releases Synthetic Dataset to Boost Small AI Models' STEM Reasoning

Tether's artificial intelligence division, Tether AI Research, has released a synthetic dataset called 'QVAC Genesis III' designed to improve the scientific, technological, engineering, and mathematical (STEM) reasoning abilities of smaller AI models. The dataset comprises 159.6 million documents and approximately 19.14 billion AI training text units (tokens). Tether reported that small models trained using this dataset achieved higher performance in multiple STEM reasoning evaluations compared to those trained on existing open-source data.
Korean Source
This article is an English localization of a Korean-language crypto news report. Original headline: 테더, 소형 AI 추론 능력 높이는 학습 데이터셋 공개