Tasks for LLMs and Their Evaluation
摘要
Since their inception, LLMs have been evaluated on a wide range of natural language tasks. These tasks include Reading Comprehension, Question Answering, Reasoning, and Text Generation. While LLMs have shown promising results, in particular as general models, their capabilities vary depending on their architecture, training dataset, and the nature of the task. We will briefly define the natural language tasks and give an overview of LLMs’ current state of the art.