neulab / ExplainaBoard

Interpretable Evaluation for AI Systems
MIT License
361 stars 36 forks source link

Add negative error #609

Closed neubig closed 1 year ago

neubig commented 1 year ago

Overview

This adds the ability to flip error-based evaluation metrics such as MAE and RMSE to be negative, so that higher values are better.

It also adds tests for MAE and RMSE, which did not exist before.

Blocked by

neubig commented 1 year ago

(Integration tests are failing but they seem to be due to an unrelated issue with eaas.)