Used for byte-level perplexity metrics in rolling loglikelihood
(cls, doc)
| 543 | |
| 544 | @classmethod |
| 545 | def count_bytes(cls, doc): |
| 546 | """Used for byte-level perplexity metrics in rolling loglikelihood""" |
| 547 | return len(doc.encode("utf-8")) |
| 548 | |
| 549 | @classmethod |
| 550 | def count_words(cls, doc): |