3 ms·
I only skimmed the paper you linked, but I don't think it says what you say it says. "We found that scaling up the model almost entirely eliminates errors in t
by ma2rten 7y ago
I only skimmed the paper you linked, but I don't think it says what you say it says.
"We found that scaling up the model almost
entirely eliminates errors in the n-gram category."
- p1esk 7y agoSee table 1 in his last link.