thanks for the codes! I have a question: should I tokenize the predictions and reference texts before using this api?

[meiling-fdu]

[riyadhrazzaq]

see the examples for a better understanding of the input format: https://github.com/Maluuba/nlg-eval/tree/master/examples