Ryan Wang

I’m a fourth-year undergraduate student studying computer science at the University of Southern California. In the past, I explored LLM memorization and contributed to developing pre-training data watermarks with statistical guarantees. I am currently advised by Robin Jia and Swabha Swayamdipta, where we’re working on methods to pre-train language models using undesired data, such as hate speech, so that the models can understand this data without learning to generate it.

In my free time, I like to sleep and eat food. If you like either one of those things, we’ll get along great! If you think I’m interesting, here’s a link (https://ryanyxw.github.io/) to my website that will be more interesting!