Leveraging robust optimization for llm alignment under distribution shifts | Read Paper on Bytez