Retrievalnormalizeddcg

retrievalnormalizeddcg

Metric RetrievalNormalizedDCG from torchmetrics (torchmetrics.RetrievalNormalizedDCG)

When to invoke this skill

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

Reference signature

from torchmetrics import RetrievalNormalizedDCG

# _RetrievalNormalizedDCG(empty_target_action: str = 'neg', ignore_index: Optional[int] = None, top_k: Optional[int] = None, **kwargs: Any) -> None

Library docstring

Wrapper for deprecated import.

>>> from torch import tensor
>>> indexes = tensor([0, 0, 0, 1, 1, 1, 1])
>>> preds = tensor([0.2, 0.3, 0.5, 0.1, 0.3, 0.5, 0.2])
>>> target = tensor([False, False, True, False, True, False, True])
>>> ndcg = _RetrievalNormalizedDCG()
>>> ndcg(preds, target, indexes=indexes)
tensor(0.8467)

Quick recipe

import torchmetrics as _m
score = _m.RetrievalNormalizedDCG(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).