def summary(self) -> DatasetSummary:
"""Generate a summary representation of this dataset.
Returns:
A summary representation of this dataset.
"""
if not self.all_fe_datasets:
print("FastEstimator-Warn: BatchDataset summary will be incomplete since non-FEDatasets were used.")
return DatasetSummary(num_instances=len(self), keys={})
summaries = [ds.summary() for ds in self.datasets]
keys = {k: v for summary in summaries for k, v in summary.keys.items()}
return DatasetSummary(num_instances=len(self), keys=keys)
def summary(self) -> DatasetSummary:
"""Generate a summary representation of this dataset.
Returns:
A summary representation of this dataset.
"""
summaries = [ds.summary() for ds in self.datasets]
keys = {k: v for summary in summaries for k, v in summary.keys.items()}
return DatasetSummary(num_instances=len(self), keys=keys)
def summary(self) -> DatasetSummary:
"""Generate a summary representation of this dataset.
Returns:
A summary representation of this dataset.
"""
sample = self[0]
key_summary = {}
for key in sample.keys():
val = sample[key]
# TODO - if val is empty list, should find a sample which has entries
shape = get_shape(val)
dtype = get_type(val)
key_summary[key] = KeySummary(num_unique_values=None,
shape=shape,
dtype=dtype)
return DatasetSummary(num_instances=self.samples_per_epoch,
keys=key_summary)