Tatsunori Hashimoto

Career

Tatsunori Hashimoto was a graduate student at MIT co-advised by Tommi Jaakkola and David Gifford, and then a postdoctoral researcher at Stanford with John C. Duchi and Percy Liang, studying trade-offs between the average and worst-case performance of machine learning models. Hashimoto is now an assistant professor of computer science at Stanford and is affiliated with the Center for Research on Foundation Models.¹⁵

Research

Hashimoto's research uses robustness and worst-case performance as a lens on problems in machine learning and NLP, including long-tail behaviour, out-of-domain understanding and fairness. A 2018 paper first-authored by Hashimoto, "Fairness Without Demographics in Repeated Loss Minimization," used distributionally robust optimization to protect minority groups without knowing their identities.¹⁶

Hashimoto has also worked on language models, co-authoring Diffusion-LM (2022) for controllable text generation, AlpacaFarm (2023), a simulator for methods that learn from human feedback, HELM (2022) and s1 (2025), a simple approach to test-time scaling. The Stanford Alpaca repository lists its authors as graduate students advised by Hashimoto.⁷²⁸³⁹

Teaching

With Percy Liang, Hashimoto teaches Stanford's CS336, Language Modeling from Scratch, which has been offered in Spring 2024, Spring 2025 and Spring 2026 and walks students through building language models end to end.⁴

Work and contributions

• HELM (Holistic Evaluation of Language Models) Published research

  Listed among the authors of the HELM paper (2022), which introduced Stanford CRFM's framework for holistic evaluation of language models.³

• Stanford Alpaca Open-source code

  The repository's README says all of its student authors were advised by Hashimoto and lists Hashimoto in the citation; the code is released under Apache 2.0, while the data and weight diff are CC BY NC 4.0.²

• CS336: Language Modeling from Scratch Published research

  Listed as an instructor, with Percy Liang, of Stanford's CS336 course in Spring 2026.⁴

Timeline

1. Jun 2018Posted "Fairness Without Demographics in Repeated Loss Minimization" as first author.⁶
2. May 2022Co-authored Diffusion-LM, a diffusion-based approach to controllable text generation.⁷
3. Nov 2022Co-authored the HELM paper on holistic evaluation of language models.³
4. May 2023Co-authored AlpacaFarm, a simulation framework for learning from human feedback.⁸
5. 2024Taught the Spring 2024 offering of CS336, Language Modeling from Scratch, at Stanford with Percy Liang.⁴¹⁰
6. Jan 2025Co-authored s1, a paper on simple test-time scaling.⁹

Where they work

• Stanford University 40 people in the wiki

Related people

• Percy LiangAlso at Stanford University
• Christopher ManningAlso at Stanford University
• Dan JurafskyAlso at Stanford University
• Rohan TaoriAlso at Stanford University
• Yoav ShohamAlso at Stanford University
• AI ExplainedTeaching and writing · Evaluation and benchmarks
• Emily M. BenderLanguage models · Teaching and writing
• Pradeep DasigiLanguage models · Evaluation and benchmarks
• Marina DanilevskyEvaluation and benchmarks · Language models
• Thomas ScialomLanguage models · Evaluation and benchmarks