PeopleofSuperintelligence

Researcher

Li Dong

Also known as 董力

Li Dong is a principal researcher in Microsoft Research Asia's General AI group, and his papers cover language-model pretraining, vision pretraining and large language models. He worked at Beihang University from 2010 and joined Microsoft Research Asia's Natural Language Computing group in 2012. His Edinburgh dissertation, Learning Natural Language Interfaces with Neural Models, was named a runner-up for the AAAI/ACM SIGAI Doctoral Dissertation Award in 2019. He co-authored Unified Language Model Pre-training for Natural Language Understanding and Generation at NeurIPS 2019, and he submitted BEiT to arXiv in June 2021. In 2023 he submitted Language Is Not All You Need and Retentive Network to arXiv, and in 2025 he received a best paper runner-up honour at an ICCV workshop.

Works
102

Publication numbers from OpenAlex, as of 15 October 2026. Citations and h-index are hidden because this OpenAlex record may be incomplete or mix in other authors.

Timeline

  1. 2010–2015

    Beihang and Microsoft Research Asia

    He works at Beihang University's State Key Laboratory of Software Development Environment from October 2010. In May 2012 he joins Microsoft Research Asia's Natural Language Computing group, with Furu Wei as mentor.

  2. 2015–2019

    Studying at Edinburgh

    He works at Edinburgh's Institute for Language, Cognition and Computation from September 2015 to July 2019.

  3. 2016

    LSTM-Networks for Machine Reading

    He co-writes Long Short-Term Memory-Networks for Machine Reading with Jianpeng Cheng and Mirella Lapata, a paper in the 2016 EMNLP proceedings.

  4. Since 2018

    A principal researcher in General AI

    He joins Microsoft Research Asia's General AI group as a principal researcher in December 2018 and still holds that role.

  5. 2019

    Unified pretraining and a runner-up

    He is one of the authors of Unified Language Model Pre-training for Natural Language Understanding and Generation, a NeurIPS 2019 paper. The 2019 AAAI/ACM SIGAI Doctoral Dissertation Award names his Edinburgh dissertation, Learning Natural Language Interfaces with Neural Models, a runner-up.

  6. June 2021

    BEiT for image transformers

    In June 2021 he submits BEiT: BERT Pre-Training of Image Transformers to arXiv.

  7. 2023

    Multimodal and retentive models

    In February 2023 he submits Language Is Not All You Need: Aligning Perception with Language Models to arXiv. In July 2023 he submits Retentive Network: A Successor to Transformer for Large Language Models to arXiv.

  8. 2025

    Best paper runner-up at ICCV

    He receives a best paper runner-up honour at the ICCV 2025 Workshop on Human-Interactive Generation and Editing.

Sources and credits

Photo: We have no published photo of Li Dong yet. Send us a free-licensed photo.

  1. Li Dong - Homepage dong.li. Checked 15 October 2026.
  2. Long Short-Term Memory-Networks for Machine Reading - ACL Anthology aclanthology.org. Checked 15 October 2026.
  3. Unified Language Model Pre-training for Natural Language Understanding and Generation proceedings.neurips.cc. Checked 15 October 2026.
  4. Doctoral Dissertation Award - SIGAI sigai.acm.org. Checked 15 October 2026.
  5. [2106.08254] BEiT: BERT Pre-Training of Image Transformers arxiv.org. Checked 15 October 2026.
  6. [2302.14045] Language Is Not All You Need: Aligning Perception with Language Models arxiv.org. Checked 15 October 2026.
  7. [2307.08621] Retentive Network: A Successor to Transformer for Large Language Models arxiv.org. Checked 15 October 2026.