no code implementations • 9 Nov 2023 • Qinyuan Ye, Maxamed Axmed, Reid Pryzant, Fereshte Khani
While recent works indicate that large language models can be meta-prompted to perform automatic prompt engineering, we argue that their potential is limited due to insufficient guidance for complex reasoning in the meta-prompt.
1 code implementation • 24 May 2023 • Qinyuan Ye, Harvey Yiyun Fu, Xiang Ren, Robin Jia
We investigate the predictability of large language model (LLM) capabilities: given records of past experiments using different model families, numbers of parameters, tasks, and numbers of in-context examples, can we accurately predict LLM performance on new experiment configurations?
1 code implementation • 24 May 2023 • Harvey Yiyun Fu, Qinyuan Ye, Albert Xu, Xiang Ren, Robin Jia
In this paper, we propose the task of ICL accuracy estimation, in which we predict the accuracy of an LLM when doing in-context learning on a new task given only unlabeled test data for that task.
1 code implementation • 25 May 2022 • Qinyuan Ye, Juan Zha, Xiang Ren
Recent works suggest that transformer models are capable of multi-tasking on diverse NLP tasks and adapting to new tasks efficiently.
1 code implementation • NAACL 2022 • Qinyuan Ye, Madian Khabsa, Mike Lewis, Sinong Wang, Xiang Ren, Aaron Jaech
Distilling state-of-the-art transformer models into lightweight student models is an effective way to reduce computation cost at inference time.
3 code implementations • EMNLP 2021 • Qinyuan Ye, Bill Yuchen Lin, Xiang Ren
Humans can learn a new language task efficiently with only few examples, by leveraging their knowledge obtained when learning prior tasks.
no code implementations • EMNLP 2021 • Qinyuan Ye, Belinda Z. Li, Sinong Wang, Benjamin Bolte, Hao Ma, Wen-tau Yih, Xiang Ren, Madian Khabsa
Current NLP models are predominantly trained through a two-stage "pre-train then fine-tune" pipeline.
1 code implementation • NeurIPS 2021 • Huihan Yao, Ying Chen, Qinyuan Ye, Xisen Jin, Xiang Ren
However, such a regularization technique lacks flexibility and coverage, since only importance scores towards a pre-defined list of features are adjusted, while more complex human knowledge such as feature interaction and pattern generalization can hardly be incorporated.
1 code implementation • ACL 2021 • Qinyuan Ye, Xiang Ren
Recent study further shows that they can learn to generalize to novel tasks, by including task descriptions as part of the source sequence and training the model with (source, target) examples.
no code implementations • 31 Dec 2020 • Qinyuan Ye, Belinda Z. Li, Sinong Wang, Benjamin Bolte, Hao Ma, Wen-tau Yih, Xiang Ren, Madian Khabsa
Thus, our policy packs task-relevant knowledge into the parameters of a language model.
2 code implementations • Findings of the Association for Computational Linguistics 2020 • Qinyuan Ye, Xiao Huang, Elizabeth Boschee, Xiang Ren
Advances in machine reading comprehension (MRC) rely heavily on the collection of large scale human-annotated examples in the form of (question, paragraph, answer) triples.
no code implementations • ACL 2020 • Dong-Ho Lee, Rahul Khanna, Bill Yuchen Lin, Jamin Chen, Seyeon Lee, Qinyuan Ye, Elizabeth Boschee, Leonardo Neves, Xiang Ren
Successfully training a deep neural network demands a huge corpus of labeled data.
1 code implementation • ICLR 2020 • Ziqi Wang, Yujia Qin, Wenxuan Zhou, Jun Yan, Qinyuan Ye, Leonardo Neves, Zhiyuan Liu, Xiang Ren
While deep neural networks have achieved impressive performance on a range of NLP tasks, these data-hungry models heavily rely on labeled data, which restricts their applications in scenarios where data annotation is expensive.
1 code implementation • IJCNLP 2019 • Qinyuan Ye, Liyuan Liu, Maosen Zhang, Xiang Ren
In this paper, we study the problem what limits the performance of DS-trained neural models, conduct thorough analyses, and identify a factor that can influence the performance greatly, shifted label distribution.