Re:33TB of text data for a 1T-parameter model
by qumpis on Sunday April 16, 2023 @06:02AM
(#35588660)
Train only on data that is hard to predict/complete?
[ Parent
| on HN ]
There are no comments on this comment yet.