Extendededitdistance

extendededitdistance

Metric ExtendedEditDistance from torchmetrics (torchmetrics.ExtendedEditDistance)

When to invoke this skill

The user has predictions + ground truth and asks to evaluate with ExtendedEditDistance, or mentions torchmetrics.ExtendedEditDistance directly, or wants the standard torchmetrics implementation.

Reference signature

from torchmetrics import ExtendedEditDistance

# _ExtendedEditDistance(language: Literal['en', 'ja'] = 'en', return_sentence_level_score: bool = False, alpha: float = 2.0, rho: float = 0.3, deletion: float = 0.2, insertion: float = 1.0, **kwargs: Any) -> None

Library docstring

Wrapper for deprecated import.

>>> preds = ["this is the prediction", "here is an other sample"]
>>> target = ["this is the reference", "here is another one"]
>>> eed = _ExtendedEditDistance()
>>> eed(preds=preds, target=target)
tensor(0.3078)

Quick recipe

import torchmetrics as _m
score = _m.ExtendedEditDistance(y_true, y_pred)

Don'ts

• Don't reimplement when the library version handles edge cases (NaN, ties, empty inputs) better than a hand-rolled formula.
• Always check the library version's argument order — sklearn is (y_true, y_pred) while torchmetrics is (preds, target).