arXiv · NLP· Abhinav Khattar (NVIDIA), Aishwarya Chhabra (NVIDIA), Ameya Velingker (NVIDIA), Anusha Kamath (NVIDIA), Brandon Norick (NVIDIA), Kanishk Singla (NVIDIA), Kiran Praveen (NVIDIA), Markus Kliegl (NVIDIA), Mohammad Qazim Bhat (NVIDIA), Mostofa Patwary (NVIDIA), Niket Agarwal (NVIDIA), Rajpreet Thethy (NVIDIA), Ramani Akkati (NVIDIA), Raviraj Joshi (NVIDIA), Sai Nishanth Padala (NVIDIA), Shuoyang Ding (NVIDIA), Sriharsha Niverty (NVIDIA), Utkarsh Vaidya (NVIDIA), Vitaly Lavrukhin (NVIDIA), Yev Meyer (NVIDIA), Anoop Kunchukuttan (AI4Bharat), Mitesh Khapra (AI4Bharat), Mohammed Safi Ur Rahman Khan (AI4Bharat), Rudra Murthy (AI4Bharat), Sumanth Doddapaneni (AI4Bharat), Thanmay Jayakumar (AI4Bharat), Ujjwal Sharma (AI4Bharat)·· 4 小时前AI 评分27
NVIDIA 与 AI4Bharat 发布 Nemotron-Pretraining-Indic-v1 合成语料库及评测套件
Scaling Synthetic Pre-training Data and Evaluation for Indic LLMs
AI 导读
NVIDIA 与 AI4Bharat 联合推出 Nemotron-Pretraining-Indic-v1 及评测套件，构建了覆盖 22 种印度语言的 2.83 万亿 token 合成预训练语料库。
来源：arXiv · NLP · arxiv.org