Closed neubig closed 1 year ago
For example, when analyzing sst2, we get these very wide confidence intervals.
When measuring the actual confidence interval according to a t-test calculator, we get much less wide confidence intervals.
Related discussion: https://github.com/neulab/ExplainaBoard/issues/510
For example, when analyzing sst2, we get these very wide confidence intervals.
When measuring the actual confidence interval according to a t-test calculator, we get much less wide confidence intervals.
Related discussion: https://github.com/neulab/ExplainaBoard/issues/510