← Search

Masih Eskandar

2 accepted papers

2025

DISCO: Disentangled Communication Steering for Large Language Models

NeurIPS 2025poster

A variety of recent methods guide large language model outputs via the inference-time addition of *steering vectors* to residual-stream or attention-head representations. In contrast, we propose to inject steering vectors directly into the query and value representation spaces within attention heads…

Cited by 0SourcecodeScholar
2025

STAR: Stability-Inducing Weight Perturbation for Continual Learning

ICLR 2025poster

Humans can naturally learn new and varying tasks in a sequential manner. Continual learning is a class of learning algorithms that updates its learned model as it sees new data (on potentially new tasks) in a sequence. A key challenge in continual learning is that as the model is updated to lea…