Zobrazeno 1 - 10
of 164
pro vyhledávání: '"Tu, Zhaopeng"'
Autor:
Yuan, Youliang, Jiao, Wenxiang, Wang, Wenxuan, Huang, Jen-tse, Xu, Jiahao, Liang, Tian, He, Pinjia, Tu, Zhaopeng
This study addresses a critical gap in safety tuning practices for Large Language Models (LLMs) by identifying and tackling a refusal position bias within safety tuning data, which compromises the models' ability to appropriately refuse generating un
Externí odkaz:
http://arxiv.org/abs/2407.09121
Existing LLMs exhibit remarkable performance on various NLP tasks, but still struggle with complex real-world tasks, even equipped with advanced strategies like CoT and ReAct. In this work, we propose the CoAct framework, which transfers the hierarch
Externí odkaz:
http://arxiv.org/abs/2406.13381
Token repetition is a typical form of multi-modal problem in fully non-autoregressive translation (NAT). In this work, we revisit the multi-modal problem in recently proposed NAT models. Our study reveals that these advanced models have introduced ot
Externí odkaz:
http://arxiv.org/abs/2405.02673
Autor:
Huang, Jen-tse, Li, Eric John, Lam, Man Ho, Liang, Tian, Wang, Wenxuan, Yuan, Youliang, Jiao, Wenxiang, Wang, Xing, Tu, Zhaopeng, Lyu, Michael R.
Decision-making is a complex process requiring diverse abilities, making it an excellent framework for evaluating Large Language Models (LLMs). Researchers have examined LLMs' decision-making through the lens of Game Theory. However, existing evaluat
Externí odkaz:
http://arxiv.org/abs/2403.11807
Autor:
He, Zhiwei, Zhou, Binglin, Hao, Hongkun, Liu, Aiwei, Wang, Xing, Tu, Zhaopeng, Zhang, Zhuosheng, Wang, Rui
Text watermarking technology aims to tag and identify content produced by large language models (LLMs) to prevent misuse. In this study, we introduce the concept of cross-lingual consistency in text watermarking, which assesses the ability of text wa
Externí odkaz:
http://arxiv.org/abs/2402.14007
Publikováno v:
CIKM (2020) 505-514
Embedding entities and relations into continuous vector spaces has attracted a surge of interest in recent years. Most embedding methods assume that all test entities are available during training, which makes it time-consuming to retrain embeddings
Externí odkaz:
http://arxiv.org/abs/2402.14033
Autor:
Guo, Zhengsheng, He, Zhiwei, Jiao, Wenxiang, Wang, Xing, Wang, Rui, Chen, Kehai, Tu, Zhaopeng, Xu, Yong, Zhang, Min
Motivated by the success of unsupervised neural machine translation (UNMT), we introduce an unsupervised sign language translation and generation network (USLNet), which learns from abundant single-modality (text and video) data without parallel sign
Externí odkaz:
http://arxiv.org/abs/2402.07726
In this paper, we re-examine the Markov property in the context of neural machine translation. We design a Markov Autoregressive Transformer~(MAT) and undertake a comprehensive assessment of its performance across four WMT benchmarks. Our findings in
Externí odkaz:
http://arxiv.org/abs/2402.02084
Autor:
Du, Cunxiao, Jiang, Jing, Yuanchen, Xu, Wu, Jiawei, Yu, Sicheng, Li, Yongqi, Li, Shenggui, Xu, Kai, Nie, Liqiang, Tu, Zhaopeng, You, Yang
Speculative decoding is a relatively new decoding framework that leverages small and efficient draft models to reduce the latency of LLMs. In this study, we introduce GliDe and CaPE, two low-hassle modifications to vanilla speculative decoding to fur
Externí odkaz:
http://arxiv.org/abs/2402.02082
Autor:
He, Zhiwei, Wang, Xing, Jiao, Wenxiang, Zhang, Zhuosheng, Wang, Rui, Shi, Shuming, Tu, Zhaopeng
Insufficient modeling of human preferences within the reward model is a major obstacle for leveraging human feedback to improve translation quality. Fortunately, quality estimation (QE), which predicts the quality of a given translation without refer
Externí odkaz:
http://arxiv.org/abs/2401.12873