FeverRoberta is a classifier model that predicts if evidence is related to query claim. The model achieved F1 score of 92.67% with test dataset "mwong/fever-evidence-related". Using pretrained roberta-base model, the classifier head is trained on Fever dataset.