Researcher

Jiacheng Liu

Also known as Gary Liu, 刘嘉程

Jiacheng Liu is a natural-language-processing researcher whose work spans commonsense reasoning, language models, and training-data search. He earned a computer science degree at the University of Illinois Urbana-Champaign and completed a PhD at the University of Washington in 2025. His papers covered generated knowledge prompting, Infini-gram, OLMoTrace, Infini-gram mini, and Olmo 3; OLMoTrace won ACL's 2025 Best Demo Paper Award, and Infini-gram mini won the EMNLP 2025 Best Paper Award. He joined xAI after his PhD and is a Member of Technical Staff at Microsoft AI.

Now at
Microsoft
Born
1994
Works
41
Citations
552
h-index
10

Years link to their sources. Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2020

    A computer science degree

    He earns a BS in computer science at the University of Illinois at Urbana-Champaign and works with Professor Julia Hockenmaier. He receives a 2020 Honorable Mention in the CRA Outstanding Undergraduate Researcher Award.

  2. 2022

    Generated knowledge prompting

    In 2022, he coauthors Generated Knowledge Prompting for Commonsense Reasoning, a method that prompts a language model to generate knowledge and adds it to the input for question answering.

  3. 2024

    Infini-gram at five trillion tokens

    In 2024, he coauthors Infini-gram, a suffix-array engine that computes arbitrary-length n-gram probabilities with millisecond-level latency after training on five trillion tokens.

  4. 2025

    A PhD at H2Lab

    He completes a PhD at the University of Washington in 2025, co-advised by Yejin Choi, and joins xAI as a Member of Technical Staff.

  5. 2025

    Tracing model outputs

    In 2025, he coauthors OLMoTrace, a system that traces language-model outputs to full, multi-trillion-token training data in real time. It wins ACL's 2025 Best Demo Paper Award.

  6. 2025

    Searching petabyte-scale text

    In 2025, he coauthors Infini-gram mini, a system for searching petabyte-scale text corpora. The paper wins the EMNLP 2025 Best Paper Award.

  7. 2025

    Olmo 3 open models

    In 2025, he coauthors Olmo 3, a fully open family of language models at 7B and 32B parameter scales.

  8. Since 2026

    Microsoft AI

    He joins Microsoft AI as a Member of Technical Staff in 2026.

Career

  1. Since 2026MicrosoftMember of Technical Staff
  2. From 2025xAIMember of Technical Staff
  3. 2021–2025Allen Institute for AI (Ai2)

Education

  1. 2025University of WashingtonPhD
  2. 2020University of Illinois Urbana-ChampaignBS

Most-cited works

  1. Generated Knowledge Prompting for Commonsense Reasoning2022 · Annual Meeting of the Association for Computational Linguistics (ACL) · 201 citationsWith Ronan Le Bras
  2. CrossWeigh: Training Named Entity Tagger from Imperfect Annotations2019 · 102 citationsWith Jiawei Han
  3. Rainier: Reinforced Knowledge Introspector for Commonsense Question Answering2022 · Conference on Empirical Methods in Natural Language Processing (EMNLP) · 35 citations
  4. Draft, Sketch, and Prove: Guiding Formal Theorem Provers with Informal Proofs2022 · 29 citations
  5. Inverse Scaling: When Bigger Isn't Better2023 · 29 citationsWith Ethan Perez and Samuel R. Bowman

Sources and credits

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

  1. Jiacheng Liu liujch1998.github.io. Checked 20 October 2026.
  2. CRA Outstanding Undergraduate Researcher Award | Siebel School of Computing and Data Science | Illinois siebelschool.illinois.edu. Checked 20 October 2026.
  3. Generated Knowledge Prompting for Commonsense Reasoning - ACL Anthology aclanthology.org. Checked 20 October 2026.
  4. Infini-gram: Scaling Unbounded n-gram Language Models to a Trillion Tokens arxiv.org. Checked 20 October 2026.
  5. People h2lab.cs.washington.edu. Checked 20 October 2026.
  6. OLMoTrace: Tracing Language Model Outputs Back to Trillions of Training Tokens - ACL Anthology aclanthology.org. Checked 20 October 2026.
  7. Allen School researchers recognized at ACL 2025 for shaping the present and future of natural language processing news.cs.washington.edu. Checked 20 October 2026.
  8. Infini-gram mini: Exact n-gram Search at the Internet Scale with FM-Index - ACL Anthology aclanthology.org. Checked 20 October 2026.
  9. Olmo 3 arxiv.org. Checked 20 October 2026.
  10. Jiacheng Liu (Microsoft AI): natural language processing lacuna.tiptreesystems.com. Checked 20 October 2026.