How to calculate the "error" in the "R@k ± error" in the result table?

[huubinh49]

The reason why I raise this issue is that my model achieves the same R@k every time I test it in the test set.