Builders & open source
Luke Zettlemoyer
Professor, Paul G. Allen School of Computer Science & Engineering, University of Washington
- Professor, Paul G. Allen School of Computer Science & Engineering, University of Washington12
- Senior Research Director, Meta FAIR, Meta1
Luke Zettlemoyer is a Professor in the Paul G. Allen School of Computer Science & Engineering at the University of Washington and a Senior Research Director at Meta FAIR. The Allen School's faculty page describes research on empirical methods for natural language semantics, new tasks and datasets, and how to train and deploy large language models, and notes that Zettlemoyer is a Fellow of the Association for Computational Linguistics. Zettlemoyer co-authored the RoBERTa pretraining study, the deep contextualized word representations paper, and QLoRA, a finetuning method that backpropagates through a frozen 4-bit quantized model into low-rank adapters so that a 65B parameter model can be finetuned on a single 48GB GPU.12345
At a glance
- Current roles
- Professor, Paul G. Allen School of Computer Science & Engineering +1 more
- Companies
- University of Washington, Meta
- Sector
- University
- Fields
- Language models, Training systems
- Education
- MIT, UW
- Work listed
- 3
- Honors
- 5
- Sources cited
- 10
- Profile reviewed
Career
Zettlemoyer is a Professor in the University of Washington's Paul G. Allen School of Computer Science & Engineering, where the faculty page notes more than ten years of teaching advanced AI and machine learning courses to undergraduate, PhD and professional master's students. Zettlemoyer is also a Senior Research Director at Meta FAIR.21
Zettlemoyer was named a Fellow of the Association for Computational Linguistics in January 2022 and was elected ACL President in December 2024, serving as vice president-elect in 2025, vice president in 2026 and president in 2027. In November 2025 Schmidt Sciences named Zettlemoyer an AI2050 Senior Fellow.1
Research
Zettlemoyer's research covers empirical methods for natural language semantics, including new learning algorithms, tasks and datasets, and, more recently, how to build self-supervision signals for pre-training large language models. Earlier work includes TriviaQA (2017), a reading comprehension dataset, and the deep contextualized word representations paper (2018).1264
Zettlemoyer co-authored a series of pretraining papers, including the RoBERTa study and BART, a denoising sequence-to-sequence model, in 2019, and Toolformer (2023), in which language models teach themselves to call external tools. The 2023 QLoRA paper introduced a method for finetuning quantized models through low-rank adapters.3785
Recent work studies mixed-modal and tokenizer-free models. Transfusion (2024) trains one model to predict text tokens and diffuse images, and the Byte Latent Transformer (2024) operates on dynamically sized byte patches instead of tokens; the BLT paper received an outstanding paper award at ACL 2025.9101
Work and contributions
QLoRA: Efficient Finetuning of Quantized LLMs Published research
Co-author with Tim Dettmers, Artidoro Pagnoni, and Ari Holtzman of the QLoRA paper, which introduces 4-bit NormalFloat, double quantization, and paged optimizers to cut finetuning memory.5
RoBERTa: A Robustly Optimized BERT Pretraining Approach Published research
Listed among the authors of the 2019 RoBERTa paper, a replication study of BERT pretraining.3
Deep contextualized word representations Published research
Co-author of the 2018 paper introducing word vectors learned from the internal states of a deep bidirectional language model.4
Timeline
- Co-authored TriviaQA, a reading comprehension dataset.6
- Co-authored "Deep contextualized word representations".4
- Co-authored the RoBERTa pretraining study.3
- Co-authored BART, a denoising sequence-to-sequence pretraining method.7
- Named a Fellow of the Association for Computational Linguistics.1
- Co-authored Toolformer, on language models that learn to use tools.8
- Co-authored QLoRA, a method for efficient finetuning of quantized language models.5
- Co-authored Transfusion, a single model for next-token prediction and image diffusion.9
- Elected President of the Association for Computational Linguistics, for the 2027 term.1
- Co-authored "Byte Latent Transformer: Patches Scale Better Than Tokens".10
- Named an AI2050 Senior Fellow by Schmidt Sciences.1
Honors and recognition
- ACL Fellow · Association for Computational Linguistics, 202212
- AI2050 Senior Fellow · Schmidt Sciences, 20251
- Outstanding Paper Award, ACL 2025 (Byte Latent Transformer) · Association for Computational Linguistics, 20251
- Presidential Early Career Award for Scientists and Engineers (PECASE) · The White House2
- Allen Distinguished Investigator Award · Paul G. Allen (Allen Distinguished Investigators program)2
Where they work
- University of Washington 19 people in the wiki
- Meta 106 people in the wiki
Related people
- Ali FarhadiAlso at University of Washington
- Carlos GuestrinAlso at University of Washington
- Xiaodong HeAlso at University of Washington
- Yejin ChoiAlso at University of Washington
- Yulia TsvetkovAlso at University of Washington
- Akshita BhagiaLanguage models · Training systems
- Liliang RenLanguage models · Training systems
- Yonghui WuLanguage models · Training systems
- Jeffrey QuesnelleLanguage models · Training systems
- Stas BekmanTraining systems · Language models
Suggest a correction or removal
Profiles cover public professional work only, and each statement cites a source. If something is out of date, missing or wrong, tell us and link a source. People can also ask for their profile to be corrected or removed.
References
10 sources · profile reviewed Oct 21, 2026 · data updated Oct 21, 2026
- Luke Zettlemoyer — homepage
- Luke Zettlemoyer — Paul G. Allen School faculty profile
- RoBERTa: A Robustly Optimized BERT Pretraining Approach (arXiv 1907.11692)
- Deep contextualized word representations (arXiv 1802.05365)
- QLoRA: Efficient Finetuning of Quantized LLMs (arXiv 2305.14314)
- TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension (arXiv 1705.03551)
- BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension (arXiv 1910.13461)
- Toolformer: Language Models Can Teach Themselves to Use Tools (arXiv 2302.04761)
- Transfusion: Predict the Next Token and Diffuse Images with One Multi-Modal Model (arXiv 2408.11039)
- Byte Latent Transformer: Patches Scale Better Than Tokens (arXiv 2412.09871)