Retrievalrecallatfixedprecision

retrievalrecallatfixedprecision

Metric RetrievalRecallAtFixedPrecision from torchmetrics (torchmetrics.RetrievalRecallAtFixedPrecision)

When to invoke this skill

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

Reference signature

from torchmetrics import RetrievalRecallAtFixedPrecision

# _RetrievalRecallAtFixedPrecision(min_precision: float = 0.0, max_k: Optional[int] = None, adaptive_k: bool = False, empty_target_action: str = 'neg', ignore_index: Optional[int] = None, **kwargs: Any) -> None

Library docstring

Wrapper for deprecated import.

>>> from torch import tensor
>>> indexes = tensor([0, 0, 0, 0, 1, 1, 1])
>>> preds = tensor([0.4, 0.01, 0.5, 0.6, 0.2, 0.3, 0.5])
>>> target = tensor([True, False, False, True, True, False, True])
>>> r = _RetrievalRecallAtFixedPrecision(min_precision=0.8)
>>> r(preds, target, indexes=indexes)
(tensor(0.5000), tensor(1))

Quick recipe

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