I am interested in many things. NLP, understanding language, and machine learning.

​

My current work focuses on open pretraining science. Model merging, babyLM, collaborative pretraining, efficient scaling laws, and (efficient) evaluation.

​

I am actively interested in other directions such as: understanding learning and (mech\any) interpretability, generalizations during training, etc.

​

I am also generally interested in Cognition, Computational Creativity and strong A.I. related fields.

​

Previously:

I was a part of Project Debater Group, studying argumentation mining and low-resource text classification.

​

My Ph.D. (HUJI) focused on using linguistic knowledge for text generation evaluation and training.​

​