SuperLim

SuperLim

Swedish NLP benchmark suite for evaluating language understanding models

Visit SuperLim

About SuperLim

SuperLim is a standardized multi-task NLP benchmark and analysis platform for evaluating Swedish natural language understanding systems, inspired by the English GLUE and SuperGLUE benchmarks. It provides a leaderboard that helps users find the best-suited Swedish language model for their specific task, while enabling model developers to compare their models across a variety of natural language understanding tasks. The benchmark suite includes tasks such as binary/ternary labelling, scoring, answer selection, and text generation, and is designed to be challenging enough that no tested model approaches ceiling performance. It was co-developed by Språkbanken Text, AI Sweden, and the National Library of Sweden (KBLab) through a Vinnova-funded project.

Resources

Product Website

Visit SuperLim's official website for product details and getting started.

Visit website →