HELM is a standardized large model evaluation framework developed by Stanford University. It aims to solve the problem of the lack of unified standards in the current evaluation of AI models through multi-dimensional quantitative analysis.