We plan to operate a leaderboard from the end of June using the KorNAT benchmark dataset.

The leaderboard will evaluate using the Social Value dataset and the Common Knowledge dataset of KorNAT.

The Social Value dataset consists of 4K, and it will be evaluated using the three metrics we proposed: SVA, A-SVA, and N-SVA. (Detailed information about the metrics can be found in the section on Social Value Alignment in 4 National Alignment Score of the paper.)

The Common Knowledge dataset consists of 6K and will be evaluated using Accuracy.