3 papers
cs.LG2026
PLATE: Plasticity-Tunable Efficient Adapters for Geometry-Aware Continual Learning
Romain Cosentino
We develop a continual learning method for pretrained models that \emph{requires no access to old-task data}, addressing a practical barrier in foundation model adaptation where pr…
cs.AI2024
Characterizing Large Language Model Geometry Helps Solve Toxicity Detection and Generation
Randall Balestriero, Romain Cosentino, Sarath Shekkizhar
Large Language Models (LLMs) drive current AI breakthroughs despite very little being known about their internal representations. In this work, we propose to shed the light on LLMs…
cs.AI2024
Reasoning in Large Language Models: A Geometric Perspective
Romain Cosentino, Sarath Shekkizhar
The advancement of large language models (LLMs) for real-world applications hinges critically on enhancing their reasoning capabilities. In this work, we explore the reasoning abil…