News for hackers. Stuff that matters.
Re:Byte Latent Transformer: Patches Scale Better Than Tokens by binarymax on Saturday December 14, 2024 @04:11PM (#42419473)
I was really excited for CANINE [1] but it never really went anywhere. Tokens are a hack. They work for the most part, but it’s clear when they don’t.

[1] https://arxiv.org/abs/2103.06874

There are no comments on this comment yet.