Large language models (LLMs) have become the foundation of many applications, leveraging their extensive capabilities in processing and understanding natural language. While many open-source LLMs have been released with technical reports, the lack of training details hinders further research and development. This paper presents the development of YuLan, a series of open-source LLMs with 12 billion parameters. The base model of YuLan is pre-trained on approximately 1.7T tokens derived from a diverse corpus, including massive English, Chinese, and multilingual texts. We design a three-stage pre-training method to enhance YuLan's overall capabilities. Subsequent phases of training incorporate instruction-tuning and human alignment, employing a substantial volume of high-quality synthesized data. To facilitate the learning of complex and long-tail knowledge, we devise a curriculum-learning framework throughout across these stages, which helps LLMs learn knowledge in an easy-to-hard manner. YuLan's training is finished on Jan, 2024 and has achieved performance on par with state-of-the-art LLMs across various English and Chinese benchmarks. This paper outlines a comprehensive technical roadmap for developing LLMs from scratch. Our model and codes are available at https://github.com/RUC-GSAI/YuLan-Chat.
@article{arxiv.2406.19853,
title = {YuLan: An Open-source Large Language Model},
author = {Yutao Zhu and Kun Zhou and Kelong Mao and Wentong Chen and Yiding Sun and Zhipeng Chen and Qian Cao and Yihan Wu and Yushuo Chen and Feng Wang and Lei Zhang and Junyi Li and Xiaolei Wang and Lei Wang and Beichen Zhang and Zican Dong and Xiaoxue Cheng and Yuhan Chen and Xinyu Tang and Yupeng Hou and Qiangqiang Ren and Xincheng Pang and Shufang Xie and Wayne Xin Zhao and Zhicheng Dou and Jiaxin Mao and Yankai Lin and Ruihua Song and Jun Xu and Xu Chen and Rui Yan and Zhewei Wei and Di Hu and Wenbing Huang and Ze-Feng Gao and Yueguo Chen and Weizheng Lu and Ji-Rong Wen},
journal= {arXiv preprint arXiv:2406.19853},
year = {2024}
}