🧑‍🔬 Jason D. Lee

🏛️ Affiliations

🎓Princeton UniversityPrinceton · United States2018–2026 · 91 paperscurrent🎓University of California, BerkeleyBerkeley · United States2015–2026 · 21 papersEducation 2015–2016 · Postdoc, Computer ScienceEmployment since 2025 · Associate Professor, EECS and Statisticscurrent🎓University of Southern CaliforniaLos Angeles · United States2016–2020 · 26 papersEmployment 2016–2019 · Assistant Professor, Departments of Data Science & Operations and Computer Science🎓Stanford UniversityStanford · United States2012–2018 · 15 papersEducation until 2015 · PhD, Computational Math🎓Duke UniversityDurham · United States2009–2012 · 3 papersEducation until 2010 · BS, Mathematics🎓University of WashingtonSeattle · United States2020–2021 · 2 papers🎓Dartmouth CollegeHanover · United States2016 · 2 papers

From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 8 more institutions on single bylines are not listed.

🧭 Fields, topics and concepts

Of its works in our library, with the number of works on each.

📅 Works by year

2020s

📄 Papers

The most cited first (OpenAlex and our library).

1. Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding Heads (2024 · 278 citations)
   Generating text with an LLM is slow because tokens come out one at a time. Each step has to read all of the model's weights from HBM, so the accelerator spends most of its time waiting on memory rather than computing. Speculative decoding helps, but it needs a separate draft model, which is costly to obtain and…
2. Fine-Tuning Language Models with Just Forward Passes (2023 · 153 citations)
   Fine-tuning large language models with backpropagation is memory-hungry. Activations, gradients and Adam's optimizer state can need roughly an order of magnitude more GPU memory than inference alone. The authors revisit zeroth-order optimization, which estimates gradients from loss differences rather than backward…
3. Teaching Arithmetic to Small Transformers (2023 · 57 citations)
   The paper asks which factors let a next-token predictor acquire arithmetic. It uses a controlled setting: small decoder-only transformers (NanoGPT and GPT-2, from 10.6M up to 124M parameters) trained from random initialization with character-level tokens. Tasks include addition, subtraction, multiplication, square…

All works

6 works of theirs in our library, the most cited first; 3 with a page here, the others link their original until we summarise them.

1. Medusa: Simple LLM Inference Acceleration Framework with Multiple Decoding HeadsarXiv:2401.10774Tianle Cai, Yuhong Li, Zhengyang Geng, et al.2024 · 278 citations
2. Fine-Tuning Language Models with Just Forward PassesConference on Neural Information Processing Systems (NeurIPS)Sadhika Malladi, Tianyu Gao, Eshaan Nichani, et al.2023 · 153 citations
3. Teaching Arithmetic to Small TransformersarXiv:2307.03381Nayoung Lee, Kartik Sreenivasan, Jason D. Lee, et al.2023 · 57 citations
4. On the Convergence and Robustness of Training GANs with Regularized Optimal Transport ↗arXiv (Cornell University) · no summary yetMaziar Sanjabi et al.2018 · 57 citations
5. Towards Understanding Hierarchical Learning: Benefits of Neural Representations ↗arXiv (Cornell University) · no summary yetMinshuo Chen et al.2020 · 17 citations
6. Metastable Dynamics of Chain-of-Thought Reasoning: Provable Benefits of Search, RL and Distillation ↗arXiv:2502.01694 · no summary yetJuno Kim et al.2025 · 6 citations

🤝 Co-authors here

9 co-authors with a page, the most papers together first.

• Danqi Chen
• Deming Chen
• Kangwook Lee
• Sadhika Malladi
• Sanjeev Arora
• Tianle Cai
• Tianyu Gao
• Tri Dao
• Yuhong Li

Venues

Where their works here were published, the most works first, with the editions of those that have a page.

• Conference on Neural Information Processing Systems (NeurIPS): NeurIPS 2023

🔗 Cited by

Papers of our library cite their works 497 times (of the 149,439 papers whose bibliographies we read); 69 citing papers with a page here →.

Updated 2 November 2026

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/authors/jason-d-lee/