All tags

Person: "sebastien-bubeck"

    not much happened today
    Shazeer et al (2024): you are overpaying for inference >13x