PeopleofSuperintelligence

Researcher

Huan Lin

Huan Lin is a natural language processing researcher whose work runs from neural machine translation to Alibaba's Qwen large language models. At Xiamen University, they were first author of a capsule-network model for multimodal machine translation in 2020 and a user-driven neural machine translation framework in 2021, both with Jinsong Su. By July 2023, they worked in the Language Technology Lab of Alibaba DAMO Academy and co-authored the multilingual PolyLM model. They co-authored the Qwen2 report in 2024, the Qwen3 Technical Report in May 2025 and the Qwen3 Embedding series from Alibaba's Tongyi Lab in June 2025. As of May 2026, they work with the Qwen Team as a contributor to Qwen-Scope, an open-source suite of sparse autoencoders for Qwen3 and Qwen3.5 models.

Timeline

  1. October 2020

    Capsule networks for multimodal translation

    At Xiamen University, they are first author of the Dynamic Context-guided Capsule Network (DCCN) for multimodal machine translation, written with Jinsong Su, Fandong Meng of Tencent WeChat AI and others. The paper appears at the 28th ACM International Conference on Multimedia (MM '20) in Seattle in October 2020.

  2. August 2021

    User-driven neural machine translation

    In August 2021, they present "Towards User-Driven Neural Machine Translation" at ACL-IJCNLP as first author, with Jinsong Su, Baosong Yang and Dayiheng Liu among the co-authors. The framework uses a cache-based module and user-driven contrastive learning to capture user traits from historical inputs, and the paper introduces UDT-Corpus, the first Chinese-English parallel corpus annotated with user behavior.

  3. 2022

    Diverse translation and keyphrase generation

    In July 2022, they publish a multi-candidate optimization framework for diverse neural machine translation as first author in Findings of NAACL 2022, training models with two reinforcement-learning strategies to balance diversity and quality. In December 2022, they co-author WR-One2Set, a keyphrase generation method presented at EMNLP in Abu Dhabi that corrects calibration errors in the One2Set paradigm.

  4. July 2023

    DAMO Academy and PolyLM

    By July 2023, they work in the Language Technology Lab of Alibaba DAMO Academy and co-author an ACL 2023 paper in Toronto that narrows the gap between upstream and downstream domains in k-nearest-neighbor machine translation. In July 2023, they also co-author PolyLM, a multilingual large language model trained on 640 billion tokens in 1.7B and 13B sizes.

  5. 2024

    Qwen2 and mGTE retrieval models

    They co-author the Qwen2 Technical Report in July 2024, which covers open-weight language models from 0.5 to 72 billion parameters. In November 2024, they co-author mGTE at EMNLP in Miami, a multilingual text representation model and reranker built on an encoder pre-trained with an 8192-token context.

  6. May 2025

    The Qwen3 Technical Report

    In May 2025, they co-author the Qwen3 Technical Report, which describes dense and mixture-of-experts models from 0.6 to 235 billion parameters. Qwen3 integrates a thinking mode for multi-step reasoning and a non-thinking mode for rapid responses into one framework, and expands multilingual support from 29 to 119 languages and dialects.

  7. June 2025

    Qwen3 Embedding at Tongyi Lab

    In June 2025, they co-author the Qwen3 Embedding series from Tongyi Lab, Alibaba Group, with embedding and reranking models in 0.6B, 4B and 8B sizes built on the Qwen3 foundation models. The series uses Qwen3 models as backbones and to synthesize training data across many domains and languages.

  8. May 2026

    Qwen-Scope sparse autoencoders

    In May 2026, they are a contributor to Qwen-Scope from the Qwen Team, an open-source suite of sparse autoencoders across 7 model variants from the Qwen3 and Qwen3.5 series. The suite turns sparse features into tools for inference-time steering, evaluation analysis, data workflows and post-training.

  1. Google had the most people shaping AI every year from 2014 to 2025Our estimates of which organizations employed the most people shaping AI from 1961 to 2025, with a year-by-year chart.
  2. The United States employs an estimated 53% of people in our data with a known employment countryOur estimates of where AI careers are concentrated now and how countries’ shares of dated positions changed over time.
  3. 46% of living people in our data with known birth and employment countries work outside their birth countryOur estimate of cross-border careers in AI, from country of birth and the country of each current organization.
  4. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionOur estimates of where the people shaping AI studied and where their careers led, from education records in cited biographies.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

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

  1. [2009.02016] Dynamic Context-guided Capsule Network for Multimodal Machine Translation ar5iv.labs.arxiv.org. Checked 19 October 2026.
  2. Towards User-Driven Neural Machine Translation - ACL Anthology aclanthology.org. Checked 19 October 2026.
  3. Bridging the Gap between Training and Inference: Multi-Candidate Optimization for Diverse Neural Machine Translation - ACL Anthology aclanthology.org. Checked 19 October 2026.
  4. WR-One2Set: Towards Well-Calibrated Keyphrase Generation - ACL Anthology aclanthology.org. Checked 19 October 2026.
  5. Bridging the Domain Gaps in Context Representations for k-Nearest Neighbor Neural Machine Translation - ACL Anthology aclanthology.org. Checked 19 October 2026.
  6. [2305.16599] Bridging the Domain Gaps in Context Representations for k-Nearest Neighbor Neural Machine Translation ar5iv.labs.arxiv.org. Checked 19 October 2026.
  7. [2307.06018] PolyLM: An Open Source Polyglot Large Language Model arxiv.org. Checked 19 October 2026.
  8. [2407.10671] Qwen2 Technical Report arxiv.org. Checked 19 October 2026.
  9. mGTE: Generalized Long-Context Text Representation and Reranking Models for Multilingual Text Retrieval - ACL Anthology aclanthology.org. Checked 19 October 2026.
  10. [2505.09388] Qwen3 Technical Report arxiv.org. Checked 19 October 2026.
  11. Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models arxiv.org. Checked 19 October 2026.
  12. Paper page - Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models huggingface.co. Checked 19 October 2026.
  13. [2605.11887] Qwen-Scope: Turning Sparse Features into Development Tools for Large Language Models arxiv.org. Checked 19 October 2026.
  14. Qwen-Scope: Turning Sparse Features into Development Tools for Large Language Models arxiv.org. Checked 19 October 2026.