自主驱动的语言模型从零开始的最小人工监督自我对齐（英文版）

英文标题：Principle-Driven Self-Alignment of Language Models from Scratch with Minimal Human Supervision中文摘要：研究提出了 SELF-ALIGN 方法，利用少量人工监督和结合原理驱动推理和 LLM 的生成能力，实现 AI 助手的自我对齐，减少人工监督的依赖，获得更好的性能，开发了 Dromedary AI 助手。英文摘要：Recent AI-assistant agents, such as ChatGPT, predominantly rely on supervisedfine-tuning (SFT) with human annotations and reinforcement learning from humanfeedback (RLHF) to align the output of large language models (LLMs) with humanintentions, ensuring they are helpful, ethical, and reli