Saumya Goyal is a second year MS student in the CS department specialising in AI. He is interested in LLMs, training techniques and learning theory. In particular, he likes thinking about how training techniques and modelling affect neural network behaviour and the theoretical limits we can reach. Recently he has been working on investigating differences in performance between pre-training and meta learning for LLMs. He completed his undergraduation from IIT Bombay in CS and likes to play tennis and dance during his free time.