Computes a safe mean of the losses. Args: losses: `Tensor` whose elements contain individual loss measurements. num_present: The number of measurable elements in `losses`. Returns: A scalar representing the mean of `losses`. If `num_present` is zero, then zero is returned.
(losses, num_present)
| 35 | |
| 36 | |
| 37 | def _safe_mean(losses, num_present): |
| 38 | """Computes a safe mean of the losses. |
| 39 | |
| 40 | Args: |
| 41 | losses: `Tensor` whose elements contain individual loss measurements. |
| 42 | num_present: The number of measurable elements in `losses`. |
| 43 | |
| 44 | Returns: |
| 45 | A scalar representing the mean of `losses`. If `num_present` is zero, |
| 46 | then zero is returned. |
| 47 | """ |
| 48 | total_loss = math_ops.reduce_sum(losses) |
| 49 | return math_ops.div_no_nan(total_loss, num_present, name='value') |
| 50 | |
| 51 | |
| 52 | def _num_elements(losses): |
no test coverage detected