Alignment_arxiv

Our new paper “Alignment Makes Language Models Normative, Not Descriptive” is uploaded to arXiv.