qqp

qqp

QQPEval(split: str = 'validation')

Bases: RolloutEvaluation

QQP evaluation using validation split.

max_new_tokens(inference: Any) -> int

Only need 1 token for yes/no answer.