AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
We Know How Machines Learn but We Don't Know How to Teach Machines
In talking to folks about data centrici my first few conversations, i've been asking folks to kind of define it. And i believe now we know how machines learn, but we don't know how to teach them. So a question is really like, how do we find this trade off? Thereis the scaling laws, which are very interesting rigts. You know, there's some deputy scaling law versus the chinchila law. Nhoth stuff we study in the hyperpermetrix experiments are all clean, but we have limited computation. We can't try all the communator explosion oofal pemeters and then in the end, make assumptions,. Then you have