Skywork: A More Open Bilingual Foundation Model

Wei Tianwen, zhaoliang, lichang zhang, Bo Zhu, Lijie Wang, yanghaihua, Biye Li, cheng cheng, Weiwei Lü, Rui Hu, lichenxia, Liu Yang, Xilin Luo, Xuejie Wu, Lunan LIU, Wenjun Cheng, CPFLAME, Jianhao Zhang, Xiaoyu Zhang, Lei Lin, wangxiaokun, Yutuan Ma, Chuanhai Dong, Yanqi Sun, Yifu Chen, Yongyi Peng, Xiaojuan Liang, Shuicheng Yan, Han Fang, Yahui Zhou

Skywork: A More Open Bilingual Foundation Model: 6 upvotes on Hugging Face Daily Papers, #9 of 10 papers on 2023-10-31. Day-by-day upvote history.

In this technical report, we present Skywork-13B, a family of large language models (LLMs) trained on a corpus of over 3.2 trillion tokens drawn from both English and Chinese texts. This bilingual foundation model is the most extensively trained and openly published LLMs of comparable size to date. We introduce a two-stage training methodology using a segmented corpus, targeting general purpose training and then domain-specific enhancement training, respectively. We show that our model not only excels on popular benchmarks, but also achieves state of the art performance in Chinese language modeling on diverse domains. Furthermore, we propose a novel leakage detection method, demonstrating that test data contamination is a pressing issue warranting further investigation by the LLM community. To spur future research, we release Skywork-13B along with checkpoints obtained during intermediate stages of the training process. We are also releasing part of our SkyPile corpus, a collection of over 150 billion tokens of web text, which is the largest high quality open Chinese pre-training corpus to date. We hope Skywork-13B and our open corpus will serve as a valuable open-source resource to democratize access to high-quality LLMs.

Paper page on Hugging Face · arXiv

Data: hysts-bot-data/daily-papers-stats and the Daily Papers API. Open data: tardellirs/paper-pulse-data. Sister project: Model Pulse, the download history of every model on the Hub.