from torch import rand, randint
from torchmetrics.classification import BinaryPrecisionAtFixedRecall
metric = BinaryPrecisionAtFixedRecall(min_recall=0.5)
metric.update(rand(10), randint(2,(10,)))
fig_, ax_ = metric.plot()  # the returned plot only shows the maximum recall value by default