1 code implementation • 16 Apr 2024 • Pengyu Cheng, Tianhao Hu, Han Xu, Zhisong Zhang, Yong Dai, Lei Han, Nan Du
Hence, we are curious about whether LLMs' reasoning ability can be further enhanced by Self-Play in this Adversarial language Game (SPAG).
1 code implementation • 12 Dec 2023 • Dun Zeng, Yong Dai, Pengyu Cheng, Longyue Wang, Tianhao Hu, Wanshun Chen, Nan Du, Zenglin Xu
Our analysis reveals a correlation between the calibration performance of reward models (RMs) and the alignment performance of LLMs.
1 code implementation • 14 Nov 2023 • Pengyu Cheng, Yifan Yang, Jian Li, Yong Dai, Tianhao Hu, Peixin Cao, Nan Du
Human preference alignment is essential to improve the interaction quality of large language models (LLMs).
1 code implementation • 6 Sep 2023 • Pengyu Cheng, Jiawen Xie, Ke Bai, Yong Dai, Nan Du
Besides, from the perspective of data efficiency, we propose a three-stage customized RM learning scheme, then empirically verify its effectiveness on both general preference datasets and our DSP set.
no code implementations • 25 Aug 2023 • Jiawen Xie, Pengyu Cheng, Xiao Liang, Yong Dai, Nan Du
Although dominant in natural language processing, transformer-based models remain challenged by the task of long-sequence processing, because the computational cost of self-attention operations in transformers swells quadratically with the input sequence length.
no code implementations • 25 Feb 2023 • Rui Wang, Pengyu Cheng, Ricardo Henao
To improve the fairness of PLMs in text generation, we propose to minimize the mutual information between the semantics in the generated text sentences and their demographic polarity, i. e., the demographic group to which the sentence is referring.
1 code implementation • 14 Nov 2022 • Pengyu Cheng, Ruineng Li
The new span is generated via a non-autoregressive masked language model, which can better preserve the local-contextual meaning of the replaced token.
1 code implementation • 11 Mar 2022 • Shengxuan Luo, Pengyu Cheng, Sheng Yu
To improve EA with dangling entities, we propose an unsupervised method called Semi-constraint Optimal Transport for Entity Alignment in Dangling cases (SoTead).
no code implementations • 2 Mar 2022 • Pengyu Cheng, ZhenHua Ling
In this paper, we propose a method of speaker adaption with intuitive prosodic features for statistical parametric speech synthesis.
1 code implementation • ICLR 2021 • Siyang Yuan, Pengyu Cheng, Ruiyi Zhang, Weituo Hao, Zhe Gan, Lawrence Carin
Voice style transfer, also called voice conversion, seeks to modify one speaker's voice to generate speech as if it came from another (target) speaker.
no code implementations • ICLR 2021 • Pengyu Cheng, Weituo Hao, Siyang Yuan, Shijing Si, Lawrence Carin
Pretrained text encoders, such as BERT, have been applied increasingly in various natural language processing (NLP) tasks, and have recently demonstrated significant performance gains.
no code implementations • 13 Aug 2020 • Weituo Hao, Nikhil Mehta, Kevin J Liang, Pengyu Cheng, Mostafa El-Khamy, Lawrence Carin
Experiments on MNIST, FashionMNIST, and CIFAR-10 demonstrate WAFFLe's significant improvement to local test performance and fairness while simultaneously providing an extra layer of security.
2 code implementations • ICML 2020 • Pengyu Cheng, Weituo Hao, Shuyang Dai, Jiachang Liu, Zhe Gan, Lawrence Carin
In this paper, we propose a novel Contrastive Log-ratio Upper Bound (CLUB) of mutual information.
no code implementations • ACL 2020 • Pengyu Cheng, Martin Renqiang Min, Dinghan Shen, Christopher Malon, Yizhe Zhang, Yitong Li, Lawrence Carin
Learning disentangled representations of natural language is essential for many NLP tasks, e. g., conditional text generation, style transfer, personalized dialogue systems, etc.
1 code implementation • 5 Oct 2019 • Pengyu Cheng, Yitong Li, Xinyuan Zhang, Liqun Cheng, David Carlson, Lawrence Carin
The relative importance of global versus local structure for the embeddings is learned automatically.
no code implementations • 5 Oct 2019 • Pengyu Cheng, Chang Liu, Chunyuan Li, Dinghan Shen, Ricardo Henao, Lawrence Carin
The Straight-Through (ST) estimator is a widely used technique for back-propagating gradients through discrete random variables.
1 code implementation • ACL 2019 • Dinghan Shen, Pengyu Cheng, Dhanasekar Sundararaman, Xinyuan Zhang, Qian Yang, Meng Tang, Asli Celikyilmaz, Lawrence Carin
Vector representations of sentences, trained on massive text corpora, are widely used as generic sentence embeddings across a variety of NLP problems.
no code implementations • ACL 2019 • Liqun Chen, Guoyin Wang, Chenyang Tao, Dinghan Shen, Pengyu Cheng, Xinyuan Zhang, Wenlin Wang, Yizhe Zhang, Lawrence Carin
Constituting highly informative network embeddings is an important tool for network analysis.
1 code implementation • 4 Jul 2018 • Chang Liu, Jingwei Zhuo, Pengyu Cheng, Ruiyi Zhang, Jun Zhu, Lawrence Carin
Particle-based variational inference methods (ParVIs) have gained attention in the Bayesian inference literature, for their capacity to yield flexible and accurate approximations.