noggin.logger.LiveMetric.to_dict

LiveMetric.to_dict() → Dict[str, numpy.ndarray]

Returns the batch data, epoch domain, and epoch data in a dictionary.

Additionally, running statistics are included in order to preserve the state of the metric.

Returns:

Dict[str, ndarray]

Notes

The encoded dictionary stores:

'batch_data' -> ndarray, shape-(N,)
'epoch_data' -> ndarray, shape-(M,)
'epoch_domain' -> ndarray, shape-(M,)
'cnt_since_epoch' -> int
'total_weighting' -> float
'running_weighted_sum' -> float
'name' -> str