52B To 1T: Lessons Learned Via Tele-flm Series

Li Xiang, Yao Yiqun, Jiang Xin, Fang Xuezhi, Wang Chao, Liu Xinzhang, Wang Zihan, Zhao Yu, Wang Xin, Huang Yuyao, Song Shuangyong, Li Yongxiang, Zhang Zheng, Zhao Bo, Sun Aixin, Wang Yequan, He Zhongjiang, Wang Zhongyuan, Li Xuelong, Huang Tiejun. Arxiv 2024

[Paper]    
Efficiency And Optimization Fine Tuning Large Scale Training Model Architecture Pretraining Methods Reinforcement Learning Scaling Laws Training Techniques

Large Language Models (LLMs) represent a significant stride toward Artificial General Intelligence. As scaling laws underscore the potential of increasing model sizes, the academic community has intensified its investigations into LLMs with capacities exceeding 50 billion parameters. This technical report builds on our prior work with Tele-FLM (also known as FLM-2), a publicly available 52-billion-parameter model. We delve into two primary areas: we first discuss our observation of Supervised Fine-tuning (SFT) on Tele-FLM-52B, which supports the “less is more” approach for SFT data construction; second, we demonstrate our experiments and analyses on the best practices for progressively growing a model from 52 billion to 102 billion, and subsequently to 1 trillion parameters. We will open-source a 1T model checkpoint, namely Tele-FLM-1T, to advance further training and research.

Similar Work