The Perils of Optimizing Learned Reward Functions: Low Training Error Does Not Guarantee Low Regret

Add code
Jun 22, 2024

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: