The author argued that using large language models directly as classifiers is flawed because they lack built-in mechanisms for handling different population distributions or calibration.
register or log in to get more services such as weather report, extended news articles and web search