Sheryl Hsu

I’m currently an undergrad at Stanford University. I have a strong engineering background and am currently passionate about ML research and startups. I'm broadly interested in generative models, reinforcement learning, and robotics. Currently, I’m working on reinforcement learning and LLMs at IRIS lab.

Some things I’ve worked on:

Better aligning LLMs with humans using different forms of preference data as part of IRIS lab
Building out and optimizing fine-tuning as a service at Foundry
Identifying malicious and vulnerable extensions at Stanford’s Emperical Security Research Group
Data visualization tools at Two Sigma
iOS app for jazz musicians (7k downloads) called elicks

News:

10/30/24

We officially released "Grounding by Trying: LLMs with Reinforcement Learning-Enhanced Retrieval" describing a reinforcement-learning based method to improve LLMs ability to use retrieval search tools.

06/24/24

Our paper "What is in the Chrome Web Store" was covered by many news outlets, notably by Forbes and The Register.

06/17/24

It’s my first day at Foundry

05/01/24

Our paper “RLVF: Learning from Verbal Feedback without Overgeneralization” was accepted to ICML

04/08/24

Stanford AI Club, which I co-founded, officially launches

01/02/24

First day of my internship at Hudson River Trading

11/15/23

Our paper “What is in the Chrome Web Store” has been accepted to AsiaCCS

06/19/23

First day of my internship at Two Sigma

09/26/22

My app elicks is officially available in the Apple App Store

Two truths and a lie:

I’ve gone stand up paddle boarding inside a drain pipe off a river
I fell off the risers (and into the trumpet section) at a band performance in Carnegie hall
I once made the entire school evacuate by improperly using a laser cutter

Contact me: