NLP

A collection of different metrics for NLP models.

BLEU(max_order, matches_by_order, ...)	Computes the BLEU score for sequence generation.
Perplexity(aggregate_crossentropy, num_samples)	Computes perplexity for sequence generation.
RougeL(total_precision, total_recall, ...)	Computes macro-averaged ROUGE-L recall, precision, and F1-score.
RougeN(total_precision, total_recall, ...)	Computes macro-averaged ROUGE-N recall, precision, and F1-score.
WER(total, count)	Computes Word Error Rate (WER) for speech recognition or text generation tasks.