AI.info

Violet Xiang

Explore Violet Xiang on AI.info.

• One Bias After Another: Mechanistic Reward Shaping and Persistent Biases in Language Reward Models