什么是后训练？大语言模型训练后优化方法综述（英文版）

The emergence of Large Language Models (LLMs) has fundamentally transformed natu-ral language processing, making them indispensable across domains ranging from conver-sational systems to scientific exploration. However, their pre-trained architectures oftenreveal limitations in specialized contexts, including restricted reasoning capacities, ethi-cal uncertainties, and suboptimal domain-specific performance. These challenges necessi-tate advanced post-training language models (PoLMs) to addre