Francesco Capuano

TLDR: A technical blog to revisit the fundamentals of what, in the crudest sense, makes Deep Learning work. A SGD-to-Muon tour, derived from first principles in math and then implemented from scratch in Jax.

TLDR: I just started my PhD at Oxford, and I could not even imagine this would become real over the last few years, when for the majority of the time I truly acted without a plan. I do believe I am someone who stumbled forward with help, luck, and all the hard work I was capable of. Now, I hope to support other hollowbreds finding their path too. Read more

TLDR: Privilege sounds like “You work too much” because the majority of ordinary people work to make ends meet, regardless of balance. Outside of the—highly privileged—tech bubble, people live very different lives, and it’s hard to change when making the switch. There are hard-wired habits related to one’s own life values.