Kobe: I agree with you actually was quite a bit, but there was recently a paper that attempted to disagree with you. Training is also just like a dark art versus inference, right? Inference is a very well defined algorithm. training involves like, I do batch norm on a random layer. And then suddenly truncated a bunch of other things. You know, there's like so much dark art in that stuff and even if you were able to prove that you could do it for one model, it'll never scale to the way people like build models.

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