Mechanistic Interpretability of GPT-like Models on Summarization Tasks

Anurag Mishra

Mechanistic interpretability research seeks to reveal the inner workings of large language models, yet most work focuses on classification or generative tasks rather than summariza…