Tianyu Gao
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Tianyu Gao has a total of 26 co-authors. Here are the top co-authors by number of shared publications:
- Di Cheng
School of Optical and Electronic Information, Suzhou Key Laboratory of Biophotonics, Suzhou City University, Suzhou, P. R. China
1 shared publication - Fan Liu
Key Laboratory of Arable Land Conservation (Middle and Lower Reaches of Yangtse River), Ministry of Agriculture and Rural affairs, College of Resources and Environment, Huazhong Agricultural University, Wuhan 430070, China
1 shared publication - Guohong Qiu
College of Resources and Environment, Huazhong Agricultural University
1 shared publication - Huaixia Yang
Pharmacy College, Henan University of Chinese Medicine, Zhengzhou, China
1 shared publication Jason Weston
1 shared publication- Jing Xu
School of Science and Technology, Harbin Institute of Technology, Shenzhen, Guangdong, China
1 shared publication
Latest Publications
- Open Access
The CRINGE Loss: Learning what language not to model
Standard language model training employs gold human documents or human-human interaction data, and treats all training data as positive examples. Growing evidence shows that even with very large amounts of positive training data, issues remain that can be alleviated with relatively small amounts of negative data -- examples of what the model should not do. In this work, we propose a novel procedure to train with such data called the CRINGE loss (ContRastive Iterative Negative GEneration). We show the effectiveness of this approach across three different experiments on the tasks of safe generation, contradiction avoidance, and open-domain dialogue. Our models outperform multiple strong baselines and are conceptually simple, easy to train and implement.
Open AccessThe CRINGE Loss: Learning what language not to model
Standard language model training employs gold human documents or human-human interaction data, and treats all training data as positive examples. Growing evidence shows that even with very large amounts of positive training data, issues remain that can be alleviated with relatively small amounts of negative data -- examples of what the model should not do. In this work, we propose a novel procedure to train with such data called the CRINGE loss (ContRastive Iterative Negative GEneration). We show the effectiveness of this approach across three different experiments on the tasks of safe generation, contradiction avoidance, and open-domain dialogue. Our models outperform multiple strong baselines and are conceptually simple, easy to train and implement.
- 1 January 2022
- Published by Optica Publishing Group
- p. M4B.5
Novel Mirror-flipped Mode Permutation Technique for Long-haul Mode-division Multiplexing Transmissions
We propose a mirror-flipped mode permutation scheme based on newly developed differential-mode-delay-symmetric fibers, which further suppresses modal-dispersion impact by 30% and mitigates mode-dependent-loss impact with 3.8-dB Q-factor improvement compared with the traditional permutation method in 1028-km 6-mode transmission.
- 1 January 2022
- Published by Optica Publishing Group
- p. M4B.5
Novel Mirror-flipped Mode Permutation Technique for Long-haul Mode-division Multiplexing Transmissions
We propose a mirror-flipped mode permutation scheme based on newly developed differential-mode-delay-symmetric fibers, which further suppresses modal-dispersion impact by 30% and mitigates mode-dependent-loss impact with 3.8-dB Q-factor improvement compared with the traditional permutation method in 1028-km 6-mode transmission.
- 1 June 2015
- Published by Elsevier in Journal of Hazardous Materials
- Vol. 290, 106-116
Oxidation process of dissolvable sulfide by synthesized todorokite in aqueous systems
- 1 June 2015
- Published by Elsevier in Journal of Hazardous Materials
- Vol. 290, 106-116
Oxidation process of dissolvable sulfide by synthesized todorokite in aqueous systems