Every story and comment from the Hacker News API, in a layout borrowed from 2001.
A comment on Byte Latent Transformer: Patches Scale Better Than Tokens. Parent comment
The point is that it can model any sequence of bytes. It's what-follows-what that matters, not how we're encoding it.
There are no comments on this comment yet.
The point is that it can model any sequence of bytes. It's what-follows-what that matters, not how we're encoding it.