Recent advancements in large language models (LLMs) have demonstrated remarkable abilities in handling a variety of natural language processing (NLP) downstream tasks, even on mathematical tasks requiring multi-step reasoning. In this report, we introduce the KwaiYiiMath which enhances the mathematical reasoning abilities of KwaiYiiBase1, by applying Supervised Fine-Tuning (SFT) and Reinforced Learning from Human Feedback (RLHF), including on both English and Chinese mathematical tasks. Meanwhile, we also constructed a small-scale Chinese primary school mathematics test set (named KMath), consisting of 188 examples to evaluate the correctness of the problem-solving process generated by the models. Empirical studies demonstrate that KwaiYiiMath can achieve state-of-the-art (SOTA) performance on GSM8k, CMath, and KMath compared with the similar size models, respectively.
@article{arxiv.2310.07488,
title = {KwaiYiiMath: Technical Report},
author = {Jiayi Fu and Lei Lin and Xiaoyang Gao and Pengli Liu and Zhengzong Chen and Zhirui Yang and Shengnan Zhang and Xue Zheng and Yan Li and Yuliang Liu and Xucheng Ye and Yiqiao Liao and Chao Liao and Bin Chen and Chengru Song and Junchen Wan and Zijia Lin and Fuzheng Zhang and Zhongyuan Wang and Di Zhang and Kun Gai},
journal= {arXiv preprint arXiv:2310.07488},
year = {2023}
}
Comments
technical report. arXiv admin note: text overlap with arXiv:2306.16636 by other authors