Mengyu Ye

I am currently a 1st year Ph.D. student in NLP at Fundamental AI lab (member of Tohoku NLP Group), Tohoku University, advised by Prof. Jun Suzuki.

My research interests lie in Natural Language Processing and Machine Learning. My work is driven by the goal of creating neural models that are fully controllable like mechanistic systems, through a thorough understanding of the algorithms they have learned. I am also interested in exploring ways to evaluate language models.

My former work focused on evaluating LLMs’ reasoning ability, conducting experiments across dozens of LLMs on multiple tasks. I am currently focusing on analyzing LLMs using Sparse autoencoders.