Zaixiang Zheng (郑在翔)   Research Scientist | ByteDance Research Email: zhengzaixiang [at] bytedance.com or zaixiang.zheng [at] gmail.com > Please don't hesitate to reach out to my ByteDance email if I don't get back to you quickly elsewhere (e.g., Github issues or Gmail).

About Me

I am a research scientist at ByteDance Research, affiliated with AI Drug Discovery Team led by Prof. Quanquan Gu.

My general research interest lies in deep probabilistic modeling and large-scale generative learning (e.g., LLMs & diffusion models) along with their applications in wide spectrum of real-world problems, especially in human language and AI for Science (e.g., generative protein modeling & design).

Before joining ByteDance, I completed my five-year PhD program in computer science at NJUNLP group, Nanjing University (09/2016-06/2021), advised by Prof. Jiajun Chen and Prof. Shujian Huang. During my PhD, I spent wonderful time at ILCC, University of Edinburgh working with Prof. Alexandra Birch (09/2019-08/2020). I also interned at MLNLC group, ByteDance AI Lab as a research intern working with Prof. Hao Zhou and Prof. Lei Li (09/2020-06/2021). I was the recipient of Chinese Information Processing Society of China (CIPS) best doctoral dissertation award, and received ACL 2021 best paper and INLG 2022 best short-paper awards, and was the leading contributor in achieving the first place in WMT 2021 German-English translation shared task with non-autoregressive translation system.

[Hiring!] We are looking for highly-motivated interns from CS/ML background. Feel free to DM me if you are interested in!

News

• Jan. 2024: I am serving as Area Chair for ACL 2024.
• Dec. 2023: I am serving as Area Chair for NAACL 2024.
• Nov. 2023: gave an invited talk about LM-DESIGN on ML for protein engineering seminar.
• Oct. 2023: I am serving as Area Chair for EACL 2024.
• May-Oct. 2023: gave an invited talk about deep generative sequence modeling for human languages and proteins at UC Santa Barbara, Tongji University, TechBeat, MLNLP seminar, IWNLG and Shanghai Uni. of Finance & Economics.
• Apr. 2023: our paper Deep Equilibrium Non-autoregressive Sequence Learning was accepted to Findings of ACL 2023!
• Apr. 2023: our LM-DESIGN paper about steering protein LLMs for designing protein sequences for desired folds was accepted to ICML 2023 as oral presentation!
• Nov. 2022: received Chinese Information Processing Society of China (CIPS) Best Doctoral Dissertation Award!
• Oct. 2022: gave an invited talk about deep generative modeling for natural languages at FudanNLP.
• Oct. 2022: one paper about multi-task learning for non-autoregressive models was accepted to EMNLP.
• July 2022: our LAFT paper about cross-lingual transfer for text generation received the Best Short-Paper Award by INLG!
• Oct. 2021: our REDER about reversible machine translation was accepted to NeurIPS 2021!
• July 2021: I joined ByteDance!
• June 2021: Finally, I passed my viva and reveived my PhD degree!
• Aug. 2020: I am looking for a full-time research scientist position in the industry.

Selected Publications/Preprints

[Google Scholar] | [^‡: equal contributions] [^#: interns/students I mentored]

• Antigen-Specific Antibody Design via Direct Energy-based Preference Optimization
  Xiangxin Zhou^‡#, Dongyu Xue^‡, Ruizhe Chen^‡, Zaixiang Zheng, Liang Wang and Quanquan Gu
  Preprint. arXiv:2403.16576, 2024
  

  [arXiv]

• Diffusion Language Models Are Versatile Protein Learners
  Xinyou Wang^‡#, Zaixiang Zheng^‡, Fei Ye, Dongyu Xue, Shujian Huang and Quanquan Gu
  Preprint. arXiv:2402.18567, 2024
  

  [arXiv]

• Diffusion Language Models Can Perform Many Tasks with Scaling and Instruction-Finetuning
  Jiasheng Ye^‡#, Zaixiang Zheng^‡, Yu Bao, Lihua Qian and Quanquan Gu
  Preprint. arXiv:2308.12219, 2023
  

  [arXiv] [Code]

• Structure-informed Language Models Are Protein Designers
  Zaixiang Zheng^‡, Yifan Deng^‡#, Dongyu Xue, Yi Zhou, Fei Ye and Quanquan Gu
  International Conference on Machine Learning (ICML), 2023
  Oral presentation
  

  [arXiv (revision ver.2)] [Code]

• DINOISER: Diffused Conditional Sequence Learning by Manipulating Noises
  Jiasheng Ye^#, Zaixiang Zheng, Yu Bao, Lihua Qian and Mingxuan Wang
  Preprint. arXiv:2302.10025, 2023
  

  [arXiv] [Code]

• Deep Equilibrium Non-autoregressive Sequence Learning
  Zaixiang Zheng, Yi Zhou and Hao Zhou
  Association for Computational Linguistics (Findings of ACL), 2023
  

  [Paper]

• LAFT: Cross-lingual Transfer for Text Generation by Language-Agnostic Finetuning
  Xianze Wu^#, Zaixiang Zheng, Hao Zhou and Yong Yu
  International Natural Language Generation Conference (INLG), 2022
  Best short-paper award
  

  [Paper]

• The Volctrans GLAT System: Non-autoregressive Translation Meets WMT21
  Lihua Qian^‡, Yi Zhou^‡, Zaixiang Zheng^‡, Yaoming Zhu, Zehui Lin, Jiangtao Feng, Shanbo Cheng, Lei Li, Mingxuan Wang and Hao Zhou
  Sixth Conference on Machine Translation (WMT'21), 2021
  Our parallel translation systems ranked #1 on German-to-English shared task, beating all strong autoregressive counterparts.
  

  [arXiv] [Code]

• Duplex Sequence-to-Sequence Learning for Reversible Machine Translation
  Zaixiang Zheng, Hao Zhou, Shujian Huang, Jiajun Chen, Jingjing Xu and Lei Li
  Neural Information Processing Systems (NeurIPS), 2021
  

  [Paper] [Code]

• Vocabulary Learning via Optimal Transport for Neural Machine Translation
  Jingjing Xu, Hao Zhou, Chun Gan, Zaixiang Zheng and Lei Li
  Association for Computational Linguistics (ACL), 2021
  Best paper award
  

  [Paper] [Code]

• Improving Self Attention Networks with Sequential Relations
  Zaixiang Zheng, Shujian Huang, Rongxiang Weng, Xin-Yu Dai and Jiajun Chen
  ACM/IEEE Transactions on Audio, Speech and Language Processing (TASLP), 2020
  

  [Paper]

• Towards Making the Most of Context in Neural Machine Translation
  Zaixiang Zheng^‡, Xiang Yue^‡#, Shujian Huang, Jiajun Chen and Alexandra Birch
  International Joint Conference on Artificial Intelligence (IJCAI), 2020
  

  [arXiv] [Code]

• Mirror-Generative Neural Machine Translation
  Zaixiang Zheng, Hao Zhou, Shujian Huang, Lei Li, Xin-Yu Dai and Jiajun Chen
  International Conference on Learning Representations (ICLR), 2020
  Highest review scores (8/8/8), long talk (oral presentation)
  

  [Paper] [Code]

• Dynamic Past and Future for Neural Machine Translation
  Zaixiang Zheng, Zhaopeng Tu, Shujian Huang, Xin-Yu Dai and Jiajun Chen
  Empirical Methods in Natural Language Processing (EMNLP), 2019
  

  [Paper] [Code]

• Modeling Past and Future for Neural Machine Translation
  Zaixiang Zheng^‡, Hao Zhou^‡, Shujian Huang, Lili Mou, Xin-Yu Dai, Jiajun Chen and Zhaopeng Tu
  Transactions of the Association of Computational Linguistics (TACL), 2018
  Presented at ACL 2018
  

  [Paper] [Code]

Academic Services

• Area Chair: EACL 2024, NAACL 2024, ACL 2024
• Conference Reviewer: ICLR/NeurIPS 2021-now, ICML 2022-now, ACL/EMNLP/NAACL 2019-2023, ARR 2021-2023, AAAI/IJCAI 2020-now
• Journal Reviewer: TPAMI 2022-now, JMLR 2024-now
• Program Committee Member of AAAI 2023 Doctoral Consortium

Misc.

• A fun fact about me: my name is pronounced like /dʒeng tsai-hsiang/ in mandarin Chinese, which could iterally be interpreted as "[I am] flying".