Re:33TB of text data for a 1T-parameter model
by Dylan16807 on Saturday April 15, 2023 @09:39PM
(#35586467)
That depends on what you mean by "accurate".
If you mean a lot of accuracy, that's obvious and doesn't really need argument. And this new fact doesn't change the argument.
If you mean a more moderate amount of accuracy, this isn't proof either way. Human brains take in less text but they put a lot more processing into it.
If you mean a lot of accuracy, that's obvious and doesn't really need argument. And this new fact doesn't change the argument.
If you mean a more moderate amount of accuracy, this isn't proof either way. Human brains take in less text but they put a lot more processing into it.