Sotabase
Home
Researchers
Career
·
Intern
,
Apple
2025–
·
PhD in Computer Science
,
Harvard University
2022–
·
Intern
,
DeepMind
·
Bachelor of Science
,
McGill University
·
Master of Science
,
McGill University
·
PhD in Computer Science
,
McGill University
·
Intern
,
ServiceNow Research
Publications
(21)
SOAP: Improving and Stabilizing Shampoo using Adam
arXiv.org · 2024
105
cited
Echo Chamber: RL Post-training Amplifies Behaviors Learned in Pretraining
arXiv.org · 2025
85
cited
Deconstructing What Makes a Good Optimizer for Language Models
International Conference on Learning Representations · 2024
37
cited
SOAP: Improving and Stabilizing Shampoo using Adam for Language Modeling
International Conference on Learning Representations · 2025
32
cited
Feature emergence via margin maximization: case studies in algebraic tasks
International Conference on Learning Representations · 2023
28
cited
Continuous MDP Homomorphisms and Homomorphic Policy Gradient
Neural Information Processing Systems · 2022
27
cited
Distributional Scaling Laws for Emergent Capabilities
arXiv.org · 2025
8
cited
Beyond Implicit Bias: The Insignificance of SGD Noise in Online Learning
International Conference on Machine Learning · 2023
7
cited
Policy Gradient Methods in the Presence of Symmetries and State Abstractions
Journal of machine learning research · 2023
7
cited
Creating a Cooperative AI Policymaking Platform through Open Source Collaboration
arXiv.org · 2024
2
cited
A Study of Policy Gradient on a Class of Exactly Solvable Models
arXiv.org · 2020
F OURIER C IRCUITS IN N EURAL N ETWORKS : U NLOCK - ING THE P OTENTIAL OF L ARGE L ANGUAGE M ODELS IN M ATHEMATICAL R EASONING AND M ODULAR A RITH - METIC
Random Scaling for Emergent Capabilities
2025
Using cognitive models to reveal value trade-offs in language models
2025
Sotabase
Rosie Zhao | Researcher Profile | Sotabase | Sotabase