no code implementations • 14 Mar 2024 • Brandon McKinzie, Zhe Gan, Jean-Philippe Fauconnier, Sam Dodge, BoWen Zhang, Philipp Dufter, Dhruti Shah, Xianzhi Du, Futang Peng, Floris Weers, Anton Belyi, Haotian Zhang, Karanjeet Singh, Doug Kang, Ankur Jain, Hongyu Hè, Max Schwarzer, Tom Gunter, Xiang Kong, Aonan Zhang, Jianyu Wang, Chong Wang, Nan Du, Tao Lei, Sam Wiseman, Guoli Yin, Mark Lee, ZiRui Wang, Ruoming Pang, Peter Grasch, Alexander Toshev, Yinfei Yang
Through careful and comprehensive ablations of the image encoder, the vision language connector, and various pre-training data choices, we identified several crucial design lessons.
Ranked #21 on Visual Question Answering on MM-Vet
no code implementations • 19 Feb 2024 • Aiwei Liu, Haoping Bai, Zhiyun Lu, Xiang Kong, Simon Wang, Jiulong Shan, Meng Cao, Lijie Wen
In this paper, we propose a method to evaluate the response preference by using the output probabilities of response pairs under contrastive prompt pairs, which could achieve better performance on LLaMA2-7B and LLaMA2-13B compared to RLAIF.
5 code implementations • 21 Sep 2022 • Xuezhe Ma, Chunting Zhou, Xiang Kong, Junxian He, Liangke Gui, Graham Neubig, Jonathan May, Luke Zettlemoyer
The design choices in the Transformer attention mechanism, including weak inductive bias and quadratic computational complexity, have limited its application for modeling long sequences.
Ranked #1 on Long-range modeling on LRA
no code implementations • EACL 2021 • Xiang Kong, Adithya Renduchintala, James Cross, Yuqing Tang, Jiatao Gu, Xian Li
Recent work in multilingual translation advances translation quality surpassing bilingual baselines using deep transformer models with increased capacity.
1 code implementation • 9 Dec 2021 • Xiang Kong, Lu Jiang, Huiwen Chang, Han Zhang, Yuan Hao, Haifeng Gong, Irfan Essa
During inference, BLT first generates a draft layout from the input and then iteratively refines it into a high-quality layout by masking out low-confident attributes.
2 code implementations • NeurIPS 2021 • Xuezhe Ma, Xiang Kong, Sinong Wang, Chunting Zhou, Jonathan May, Hao Ma, Luke Zettlemoyer
Specifically, with the first attention function, Luna packs the input sequence into a sequence of fixed length.
1 code implementation • Findings (ACL) 2021 • Jiatao Gu, Xiang Kong
Fully non-autoregressive neural machine translation (NAT) is proposed to simultaneously predict tokens with single forward of neural networks, which significantly reduces the inference latency at the expense of quality drop compared to the Transformer baseline.
1 code implementation • EMNLP 2020 • Xiang Kong, Zhisong Zhang, Eduard Hovy
In this work, we introduce a novel local autoregressive translation (LAT) mechanism into non-autoregressive translation (NAT) models so as to capture local dependencies among tar-get outputs.
Ranked #2 on Machine Translation on WMT2016 English-Romanian
1 code implementation • Findings of the Association for Computational Linguistics 2020 • Zhisong Zhang, Xiang Kong, Lori Levin, Eduard Hovy
Recently, pre-training contextualized encoders with language model (LM) objectives has been shown an effective semi-supervised method for structured prediction.
1 code implementation • NeurIPS 2020 • Xi-An Li, Asa Cooper Stickland, Yuqing Tang, Xiang Kong
As an extension of this framework, we propose a novel method to train one shared Transformer network for multilingual machine translation with different layer selection posteriors for each language pair.
no code implementations • ACL 2020 • Zhisong Zhang, Xiang Kong, Zhengzhong Liu, Xuezhe Ma, Eduard Hovy
It remains a challenge to detect implicit arguments, calling for more future work of document-level modeling for this task.
1 code implementation • ACL 2020 • Xiang Kong, Varun Gangal, Eduard Hovy
We introduce SCDE, a dataset to evaluate the performance of computational models through sentence prediction.
Ranked #3 on Question Answering on SCDE
1 code implementation • ICLR 2021 • Xuezhe Ma, Xiang Kong, Shanghang Zhang, Eduard Hovy
In this work, we propose a new generative model that is capable of automatically decoupling global and local representations of images in an entirely unsupervised setting, by embedding a generative flow in the VAE framework to model the decoder.
1 code implementation • 11 Nov 2019 • Xiang Kong, Xianyang Chen, Eduard Hovy
Specifically, embeddings of entities and relationships are first decompressed to a more expressive and robust space by decompressing functions, then knowledge graph embedding models are trained in this new feature space.
Ranked #26 on Link Prediction on FB15k-237
no code implementations • ACL 2019 • Mengzhou Xia, Xiang Kong, Antonios Anastasopoulos, Graham Neubig
Translation to or from low-resource languages LRLs poses challenges for machine translation in terms of both adequacy and fluency.
2 code implementations • NeurIPS 2019 • Xuezhe Ma, Xiang Kong, Shanghang Zhang, Eduard Hovy
Flow-based generative models, conceptually attractive due to tractability of both the exact log-likelihood computation and latent-variable inference, and efficiency of both training and sampling, has led to a number of impressive empirical successes and spawned many advanced variants and theoretical investigations.
Ranked #3 on Image Generation on CelebA 256x256
no code implementations • 22 Jan 2019 • Xiang Kong, Bohan Li, Graham Neubig, Eduard Hovy, Yiming Yang
In this work, we propose a method for neural dialogue response generation that allows not only generating semantically reasonable responses according to the dialogue history, but also explicitly controlling the sentiment of the response via sentiment labels.
no code implementations • 21 Nov 2018 • Xiang Kong, Zhaopeng Tu, Shuming Shi, Eduard Hovy, Tong Zhang
Although Neural Machine Translation (NMT) models have advanced state-of-the-art performance in machine translation, they face problems like the inadequate translation.
Ranked #35 on Machine Translation on WMT2014 English-German
1 code implementation • 25 Sep 2018 • Xiang Kong, Qizhe Xie, Zihang Dai, Eduard Hovy
Mixture of Softmaxes (MoS) has been shown to be effective at addressing the expressiveness limitation of Softmax-based models.
Ranked #18 on Machine Translation on WMT2014 English-French
no code implementations • 13 Dec 2016 • Xiang Kong, Jeung-Yoon Choi, Stefanie Shattuck-Hufnagel
This paper describes methods for evaluating automatic speech recognition (ASR) systems in comparison with human perception results, using measures derived from linguistic distinctive features.
Automatic Speech Recognition Automatic Speech Recognition (ASR) +1
no code implementations • 13 Dec 2016 • Xiang Kong, Preethi Jyothi, Mark Hasegawa-Johnson
Mismatched transcriptions have been proposed as a mean to acquire probabilistic transcriptions from non-native speakers of a language. Prior work has demonstrated the value of these transcriptions by successfully adapting cross-lingual ASR systems for different tar-get languages.
no code implementations • 10 Nov 2016 • Xiang Kong, Xuesong Yang, Mark Hasegawa-Johnson, Jeung-Yoon Choi, Stefanie Shattuck-Hufnagel
Three consonant voicing classifiers were developed: (1) manually selected acoustic features anchored at a phonetic landmark, (2) MFCCs (either averaged across the segment or anchored at the landmark), and(3) acoustic features computed using a convolutional neural network (CNN).