lm_polygraph.stat_calculators.cross_encoder_similarity module

class lm_polygraph.stat_calculators.cross_encoder_similarity.CrossEncoderSimilarityMatrixCalculator(batch_size: int = 10, cross_encoder_name: str = 'cross-encoder/stsb-roberta-large')[source]

Bases: StatCalculator

Calculates the cross-encoder similarity matrix for generation samples using RoBERTa model.

static meta_info() → Tuple[List[str], List[str]][source]: Returns the statistics and dependencies for the calculator.