Eugene Choi

Welcome!

I am a Member of Technical Staff at Cohere, working on preference modeling and reinforcement learning for LLMs.

Previously, I was a research engineer at Reality Defender. Prior to that, I conducted machine learning research in the Machine Learning for Language (ML²) group at New York University. At ML², I had the privilege of being advised by Dr. Cheolhyoung Lee and Prof. Kyunghyun Cho. Also, I'm grateful to have closely worked with Dr. Jason Phang and Prof. Sam Bowman.

Highlights

AI Text Classifier Evaluation by Mediacorp:

An evaluation of the AI-generated text detector that I built at Reality Defender, against the detectors developed by OpenAI and GPTZero.

Academic Background

M.S. Courant Institute, New York University

Computer Science
CeDiD: 23FD-I09H-E3IW

B.A. Courant Institute, New York University

Computer Science and a minor in Mathematics
CeDiD: 22LH-QC4F-EFIF

Research

Publications and preprints (2023-)

A Non-monotonic Self-terminating Language Model

Eugene Choi, Kyunghyun Cho, Cheolhyoung Lee
11th International Conference on Learning Representations (ICLR), 2023
tl;dr: our proposed method prevents non-terminating sequences resulting from incomplete decoding algorithms
[paper] [openreview] [iclr] [code] [bibtex]

Teaching

External

African Masters of Machine Intelligence (AMMI)

Deep Learning for NLP: a modern NLP foundations class by Profs Kyunghyun Cho, Duygu Ataman
May 2022, Teaching Assistant / Lab Instructor (virtual)

At New York University

DS-UA 203: Machine Learning for Language Understanding