Hi!

I’m currently a third-year PhD student at Berkeley AI Research, advised by Anca Dragan and Dan Klein. My research is supported by the Apple Scholars in AI Fellowship.

I’m interested in building agents that can collaborate and interact with humans, and use language as a medium to do so. Currently, I’m excited about dialogue and language + RL.

Previously, I worked on research and product at Lilt, working on human-in-the-loop machine translation / Copilot for expert translators. I graduated with a double-major in computer science and philosophy from MIT, where I did research on human-inspired AI with the Computational Cognitive Science Group, advised by Kelsey Allen and Josh Tenenbaum, and machine learning security as a founding member of labsix. I also spent a great summer with the Natural Language Understanding group at Google Research NY, advised by David Weiss.

Publications

Decision-Oriented Dialogue for Human-AI Collaboration

Jessy Lin*, Nicholas Tomlin*, Jacob Andreas, Jason Eisner

We introduce a new task and suite of environments to evaluate how agents like LLMs can assist humans with everyday decision-making.

Preprint.

[Site] [Code]

InCoder: A Generative Model for Code Infilling and Synthesis

Daniel Fried*, Armen Aghajanyan*, Jessy Lin, Sida Wang, Eric Wallace, Freda Shi, Ruiqi Zhong, Wen-tau Yih, Luke Zettlemoyer, Mike Lewis

We open-source a new large language model for code that can both generate and fill-in-the-blank to do tasks like docstring generation, code rewriting, type hint inference, and more.

ICLR 2023 (oral).

[Paper] [Twitter] [Demo] [Site] [Code]

Automatic Correction of Human Translations

Jessy Lin, Geza Kovacs, Aditya Shastry, Joern Wuebker, John DeNero

We introduce the task of translation error correction and show how models can augment professional translators in-the-loop.

NAACL 2022. Best Task Paper, Best Resource Paper, Best Theme Paper Honorable Mention.

[Paper] [Twitter] [Data]

Uni[MASK]: Unified Inference in Sequential Decision Problems

Micah Carroll, Orr Paradise, Jessy Lin, Raluca Georgescu, Mingfei Sun, David Bignell, Stephanie Milani, Katja Hofmann, Matthew Hausknecht, Anca Dragan, Sam Devlin

We show how a single model trained with a BERT-like masked prediction objective can unify inference in sequential decisionmaking settings (e.g. for RL): behavior cloning, future prediction, and more.

NeurIPS 2022 (oral).

[Paper] [Twitter]

Inferring Rewards from Language in Context

Jessy Lin, Daniel Fried, Dan Klein, Anca Dragan

We infer human preferences (reward functions) from language.

ACL 2022.

[Paper] [Twitter] [Code]

Black-box Adversarial Attacks with Limited Queries and Information

Andrew Ilyas*, Logan Engstrom*, Anish Athalye*, Jessy Lin*

We generate adversarial examples for real-world ML systems like the Google Cloud Vision API using only access to predicted labels.

ICML 2018.

[Paper] [Blog] [Code]