Felix Yu works on improving Gemini's fundamental capabilities of memorizing and retrieving information and applying the capabilities in Ads, memory/personalization, long-context and more.
How media typically covers Felix Yu
Felix Yu as author
BlockRank enables scalable in-context ranking with generative models by reducing attention complexity from quadratic to linear, achieving 4.7x inference speedup while handling ~500 documents in-context within a second.
“Co-author of the paper on Scalable In-context Ranking with Generative Models”