The Perils of Optimizing Learned Reward Functions
"奖励学习" 相关文章
The Perils of Optimizing Learned Reward Functions: Low Training Error Does Not Guarantee Low Regret
Knocking Down My AI Optimist Strawman
Generalizable Reward Model (GRM): An Efficient AI Approach to Improve the Generalizability and Robustness of Reward Learning for LLMs
