Lex Fridman Podcast
#434 – Aravind Srinivas: Perplexity CEO on Future of AI, Search & the Internet
Aravind Srinivas
That's why you make the model bigger, the same RLHF on the bigger model ends up, like GPT-4 ends up making ChatGPT much better than 3.5. But that data like, oh, for this coding query, make sure the answer is formatted with these markdown and like syntax highlighting tool use and knows when to use what tools. It can decompose the query into pieces.
0
💬
0
Comments
Log in to comment.
There are no comments yet.