diff options
| author |  A. Unique TensorFlower <gardener@tensorflow.org> | 2018-08-23 12:41:07 -0700 |
|---|---|---|
| committer | TensorFlower Gardener <gardener@tensorflow.org> | 2018-08-23 12:47:37 -0700 |
| commit | cddc30aaafc35868848608cd4db40b73ff42bbfe (patch) | |
| tree | d6dae8944246f875b36db242bc4b368ecc506edf /tensorflow/python/training | |
| parent | cb00181bf7e3417b2ab64756dc0c535b5d1d1332 (diff) | |
Currently, recall_at_precision returns the threshold where the precision is
closest to the target precision, meaning the precision at the threshold can be
smaller or larger than the target precision. Some application requries the
precision to be strictly above some target precision. For such applications,
it's preferrable to return recall of 0 if we are not able to reach the target
precision. The change is compatible with existing usage.
PiperOrigin-RevId: 209981819
Diffstat (limited to 'tensorflow/python/training')
0 files changed, 0 insertions, 0 deletions