PeopleofSuperintelligence

Researcher

Yazheng Yang

Yazheng Yang is a researcher whose AI work spans language generation, multimodal models, tabular data, and speech. He coauthored a natural-language-inference paper in 2018 and TopNet for story generation in 2021. His later work includes M³IT, UniTabE, Reka's Core, Flash, and Edge report, VLFeedback, and CogAlpha for financial signal discovery. As of October 2026, he is an assistant professor at Xiamen University's Digital Media Computing Center and holds a PhD from the University of Hong Kong.

Timeline

  1. 2018

    A natural-language-inference paper

    In 2018, he coauthors a natural-language-inference paper that uses discourse markers and reinforcement learning to improve model predictions.

  2. 2021

    TopNet for story generation

    In 2021, he coauthors TopNet for long-story generation. It uses a neural topic model to sample related words as a story skeleton from a short input.

  3. 2023

    M³IT for multimodal instructions

    In 2023, he coauthors M³IT, a multimodal and multilingual instruction-tuning dataset built from 40 curated datasets and 2.4 million instances.

  4. 2024

    UniTabE for tabular data

    At ICLR 2024, he coauthors UniTabE, a pretraining protocol that processes varied table structures using TabUnit elements and a Transformer encoder.

  5. April 2024

    Reka's multimodal models

    In April 2024, he coauthors Reka's report introducing Core, Flash, and Edge, multimodal language models trained from scratch that process text, images, video, and audio.

  6. November 2024

    VLFeedback for model alignment

    In November 2024, he coauthors VLFeedback, a vision-language feedback dataset with over 82,000 multimodal instructions and rationales generated by off-the-shelf models. The paper introduces Silkie, a vision-language model fine-tuned on VLFeedback through direct preference optimization.

  7. 2025–July 2026

    KDD service and CogAlpha

    In 2025, he serves on KDD's Research Track Program Committee with the University of Hong Kong as his affiliation. In July 2026, he coauthors CogAlpha, a financial-signal discovery framework that combines code-level alpha representations with LLM-driven reasoning and evolutionary search.

  8. As of October 2026

    Assistant professor at Xiamen

    As of October 2026, he is an assistant professor and master's supervisor at Xiamen University's Digital Media Computing Center. He holds a PhD from the University of Hong Kong and researches natural language processing, pretraining for unstructured data, and speech language models. His 2026 work also includes WnW, a paper about a waxing-and-waning KV cache for long-form speech language models.

Career

  1. NowXiamen UniversityAssistant Professor and Master's Supervisor

Education

  1. University of Hong KongPhD

From the biography.

  1. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionIncludes Xiamen University. We estimate where the people shaping AI studied and where their careers led. Massachusetts Institute of Technology leads with 467 former students, including 291 recorded doctorates.
  2. The company share of AI researchers’ top papers was 31% in 2020–2024 and 8% in 1970–1974Includes Xiamen University. We estimate where the people shaping AI worked when their most-cited papers appeared: the company share was 31% in 2020–2024 and 8% in 1970–1974, and Google led all organizations in 2020–2024.
  3. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Xiamen University. The median h-index is 48 at universities and 14 at frontier AI labs. We compare publishing careers, recent citations, dual appointments and job moves across 2,420 university researchers and 308 frontier-lab researchers.
  4. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Xiamen University. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other university. We compare doctorate and 5+ year employment links, career timing and current workplaces for 187 major-prize and 1,111 best-paper winners.

Sources and credits

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

  1. Discourse Marker Augmented Network with Reinforcement Learning for Natural Language Inference aclanthology.org. Checked 21 October 2026.
  2. TopNet: Learning from Neural Topic Model to Generate Long Stories arxiv.org. Checked 21 October 2026.
  3. M$^3$IT: A Large-Scale Dataset towards Multi-Modal Multilingual Instruction Tuning arxiv.org. Checked 21 October 2026.
  4. UniTabE: A Universal Pretraining Protocol for Tabular Foundation Model in Data Science proceedings.iclr.cc. Checked 21 October 2026.
  5. Reka Core, Flash, and Edge: A Series of Powerful Multimodal Language Models arxiv.org. Checked 21 October 2026.
  6. VLFeedback: A Large-Scale AI Feedback Dataset for Large Vision-Language Models Alignment aclanthology.org. Checked 21 October 2026.
  7. Research Track Program Committee – KDD 2025 kdd.org. Checked 21 October 2026.
  8. Cognitive Alpha Mining via LLM-Driven Code-Based Evolution aclanthology.org. Checked 21 October 2026.
  9. 杨雅正-厦门大学数字媒体计算中心 cdmc.xmu.edu.cn. Checked 21 October 2026.