4 ms·
"The perfect outcome is text identical to the training set." Huh? If the LLM was only ever spitting back identical content straight from the training set, that
by Solvency 3y ago
"The perfect outcome is text identical to the training set."
Huh? If the LLM was only ever spitting back identical content straight from the training set, that would be a symptom of extreme overfitting, which everyone universally agrees is a bad thing — not a perfect thing.
- aezart 3y agoIt's not the most useful outcome for the end user, but it's the perfect outcome from the perspective of the learning algorithm. All these things care about is minimizing their loss function, where loss is deviation from the training set.