4 ms·
Somewhat related. GPT-4's calibration (predicted confidence in an answer vs probability of being correct) is severely tanked after the RLHF process. https://im
by famouswaffles 3y ago
Somewhat related. GPT-4's calibration (predicted confidence in an answer vs probability of being correct) is severely tanked after the RLHF process.
https://imgur.com/a/3gYel9r https://imgur.com/a/3gYel9r