Sherry Yang

This account is verified. Learn more

@sherryyangML

Assistant Professor @NYU_Courant; Staff Research Scientist @GoogleDeepMind. Previously Post-Doc @Stanford, PhD @UCBerkeley, M.Eng. / B.S. @MIT.

sherryy.github.io Joined September 2015

508 Following 5.3K Followers 294 Posts

Latest posts 5

Sherry Yang @sherryyangML Oct 22

World models go beyond robotics.

Autoresearch and agentic ML engineering face a familiar bottleneck: expensive experiments.

Introducing MLE-World: learn from past experiments to predict new outcomes before running them.

Paper, model, data, code: mle-world.github.io

Ko Brian Verified account @ko_brian23869 Oct 22

🚀 Toward RSI: AI agents automate ML research, but each step trains and evaluates a model, limiting test-time scaling.

Our new paper, MLE-World, asks: can agents explore more ideas by predicting experiment outcomes before running them? 🧵⬇️

View quoted post on X

Sherry Yang @sherryyangML Oct 19

Why should agentic robots rediscover what humans already know?

Excited to share our new work, MoCaP (mobile Code-As-Policy), which turns passive human videos of everyday life into a head start for agentic robots: where things are and how to approach them.

Human experience → less robot trial and error.

Tatsuya Kamijo Verified account @tatsukamijo Oct 17

Coding agents are emerging as general robot policies, but they are still inefficient (e.g., requiring long real-world exploration), especially in mobile manipulation.

Humans become efficient by watching others.
Can coding agents learn from passive human videos, just like we do?

Introducing MoCaP: Mobile Code-as-Policies for Zero-Shot Mobile Manipulation from Passive Human Videos

↓🧵

View quoted post on X

Sherry Yang @sherryyangML Oct 14

Excited to share Reward-Rate Policy Gradient: RL that accounts for execution time.

For autoresearch and ML engineering agents, actions can take seconds or hours. A fixed-step MDP misses this distinction.

We optimize MLE agents' reward per unit time in a semi-MDP.

arxiv.org/pdf/2609.36393
muhang-tian.com/reward-rate

See details below:

Muhang Tian Verified account @muhang_tian Oct 5

1/7
Excited to share our new work on agentic RL! 🛠️

Some tasks in agentic RL involve actions that incur costs. One example is machine learning engineering (MLE) agents, where a proposed script might run for minutes or hours. We propose Reward-rate Policy Gradient (RPG) using a Semi-Markov Decision Process formulation that optimizes reward per unit time.

🌐muhang-tian.com/reward-rate/
View quoted post on X

Sherry Yang @sherryyangML Sep 18

Robotics researchers in academia including myself might be having a bit of an existential crisis watching systems like Astra get better so fast.

But academia has an unusual advantage: robots doing real science in physical labs.

That's why I'm excited about WetRobo 🧬🤖, where coding agents observe a wet lab, write + execute robot programs, and adapt through physical trial and error.

Paper: arxiv.org/abs/2609.18435
Code: github.com/tsudalab/WetRo…

0:31

Sherry Yang @sherryyangML May 22

Excited to share CrystalReasoner, a reasoning model for crystal structure generation with LLMs and property-conditioned generation through RL:

Website: crystalreasoner.github.io
Paper: arxiv.org/abs/2605.14344
Code: github.com/wyy603/Crystal…

0:13