Re:Jev: New frontier model 40-400x cheaper and 20-200x faster
by ta988 on Tuesday September 15, 2026 @05:12PM
(#49718975)
Current models go beyond the simple embedding because you start to encode groups of concepts in the context-aware part of the model (attention heads or any other method). So it is never simply words/tokens in isolation anymore.
[ Parent
| on HN ]
There are no comments on this comment yet.