3 ms·
LLMs literally just place words one in front of another based on a probability and the "goodness" of training data. Of course it's going to make stuff up.
by throwawaaarrgh 3y ago
LLMs literally just place words one in front of another based on a probability and the "goodness" of training data. Of course it's going to make stuff up.