With silu or salu, this kind of smooth functions, you can construct like quadratic functions with just two or three neurons. But in practice, what really happens is that neural networks have two neurons. Both neurons have their own first order terms, but somehow in the later layers, they weighted them such that their first order terms cancel each other. The second order term survives in tail expansion. So it's a little bit shocking to me, like in some sense, neural networks are more clever than myself. It's my unreasonable craze for silu or Salu.

Get the Snipd
podcast app

Unlock the knowledge in podcasts with the podcast player of the future.
App store bannerPlay store banner

AI-powered
podcast player

Listen to all your favourite podcasts with AI-powered features

Discover
highlights

Listen to the best highlights from the podcasts you love and dive into the full episode

Save any
moment

Hear something you like? Tap your headphones to save it with AI-generated key takeaways

Share
& Export

Send highlights to Twitter, WhatsApp or export them to Notion, Readwise & more

AI-powered
podcast player

Listen to all your favourite podcasts with AI-powered features

Discover
highlights

Listen to the best highlights from the podcasts you love and dive into the full episode