🧑‍🔬 Fangyun Wei

Affiliations

• Microsoft Research Asia (China) · Beijing, China — current · 2019–2026 on 50 papers
• Microsoft (United States) · Redmond, United States — current · 2020–2026 on 10 papers
• The University of Sydney · Sydney, Australia — 2024–2025 on 4 papers
• Peking University · Beijing, China — 2014–2023 on 3 papers

From the bylines of their papers (OpenAlex, 2026-10-14), ORCID and Wikidata; institutions on papers are matched by machine. 3 more institutions on single bylines are not listed.

Topics

Language models & pretraining1

Papers

The most cited first (OpenAlex and our library).

1. EAGLE: Speculative Sampling Requires Rethinking Feature Uncertainty (2024 · 204 citations)
   Generating text token by token makes LLM inference slow. Speculative sampling speeds it up by letting a cheap model propose several tokens that the full model then checks in a single pass. However, good draft models are hard to find, especially for the smallest model in a family. EAGLE replaces the separate draft…
2. EAGLE-2: Faster Inference of Language Models with Dynamic Draft Trees (2024 · 131 citations)
   Speculative sampling makes LLM inference faster: a cheap draft model proposes tokens, and the original model checks them in parallel.

Co-authors here

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

• Chao Zhang ×2
• Hongyang Zhang ×2
• Yuhui Li ×2

Cited by

Papers of our library cite their works 295 times (of the 140 884 papers whose bibliographies we read); 3 citing papers with a page here →.

This page in Markdown → https://md.research.elamur.ai/free/en/authors/fangyun-wei/