OpenBA: An Open-sourced 15B Bilingual Asymmetric seq2seq Model Pre-trained from Scratch

Juntao Li, Zecheng Tang, Ding, Pinzheng Wang, Pei Guo, youwangjie, Dan Qiao, Wenliang Chen, Guohong Fu, Qiaoming Zhu, Guodong Zhou, Min Zhang

OpenBA: An Open-sourced 15B Bilingual Asymmetric seq2seq Model Pre-trained from Scratch: 17 upvotes on Hugging Face Daily Papers, #5 of 9 papers on 2023-09-20. Day-by-day upvote history.

Large language models (LLMs) with billions of parameters have demonstrated outstanding performance on various natural language processing tasks. This report presents OpenBA, an open-sourced 15B bilingual asymmetric seq2seq model, to contribute an LLM variant to the Chinese-oriented open-source model community. We enhance OpenBA with effective and efficient techniques as well as adopt a three-stage training strategy to train the model from scratch. Our solution can also achieve very competitive performance with only 380B tokens, which is better than LLaMA-70B on the BELEBELE benchmark, BLOOM-176B on the MMLU benchmark, GLM-130B on the C-Eval (hard) benchmark. This report provides the main details to pre-train an analogous model, including pre-training data processing, Bilingual Flan data collection, the empirical observations that inspire our model architecture design, training objectives of different stages, and other enhancement techniques. We have refactored our code to follow the design principles of the Huggingface Transformers Library, making it more convenient for developers to use, and released checkpoints of different training stages at https://huggingface.co/openBA. More details of our project are available at https://github.com/OpenNLG/openBA.git.

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.