"//tensorflow/python/data/ops:readers",
"//tensorflow/python/data/util:nest",
"//tensorflow/python/data/util:sparse",
- "//tensorflow/python:platform",
],
)
from tensorflow.python.ops import array_ops
from tensorflow.python.ops import math_ops
from tensorflow.python.util import deprecation
-from tensorflow.python.platform import tf_logging as logging
def parallel_interleave(map_func,
selector_input = dataset_ops.Dataset.zip(
(logits_ds, random_ops.RandomDataset(seed).batch(2))).map(select_dataset)
- logging.warn('selector_input.output_types: %s', str(selector_input.output_types))
- logging.warn('selector_input.output_shapes: %s', str(selector_input.output_shapes))
- for i, dataset in enumerate(datasets):
- logging.warn('dataset %i output_types: %s' % (i, str(dataset.output_types)))
- logging.warn('dataset %i output_shapes: %s' % (i, str(dataset.output_shapes)))
return DirectedInterleaveDataset(selector_input, datasets)