Cross-validation -- compute actual scores based on ratings

tuetschek / ratpred

Referenceless quality estimation for natural language generation – predicting human ratings

Other

8 stars 0 forks source link

Closed tuetschek closed 7 years ago

tuetschek commented 7 years ago

Averaging the computations is not completely accurate.

tuetschek commented 7 years ago

Fixed in fe7140013fd018119e41e85aac9b695fa1e015bd.

tuetschek commented 7 years ago

Completely fixed in 33a95694d35bae1309962995c93a9e9b1d7093c3.