# Aligning Large Language Models with Human: A Survey

An academic survey and paper collection on aligning large language models with human expectations, covering data collection, training methodologies and evaluation for alignment research.

Aligning Large Language Models with Human: A Survey is a research survey and accompanying collection of papers on aligning large language models with human expectations. It covers alignment data, including data collected from humans and from strong LLMs and instruction management; alignment training, including online and offline human alignment and parameter-efficient training; and alignment evaluation, including design principles, benchmarks, evaluation paradigms and existing alignment toolkits.

The repository organizes related work under these headings as a reference for researchers and practitioners studying why LLMs still misunderstand instructions, generate biased content, or produce factually incorrect output despite strong performance on standard benchmarks, and closes with directions the authors see as promising for future alignment research.

It is a paper and reference list rather than a piece of software, published alongside an arXiv preprint.

Website: https://arxiv.org/abs/2307.12966