CLUE

Why we included this project

CLUE is the benchmark most people turn to when they need a fixed, comparable footing for Chinese-language NLP models. It bundles a broad set of Chinese understanding tasks, from text classification and sentence-pair similarity to natural language inference, and maintains a public leaderboard so you can see how a candidate model performs before committing to it. The repo also includes baseline implementations and links to pre-trained Chinese models, which gives you a sensible starting point instead of training from scratch. It is a research-community standard, not a product you deploy, but for anyone selecting or sanity-checking Chinese models, the datasets and leaderboard history save real time. Because it spans older and current task families, it also works as a window into how Chinese NLU has progressed.