Skip to content

Latest commit

 

History

History
30 lines (22 loc) · 1.4 KB

bulk_fixed_feature_ids.md

File metadata and controls

30 lines (22 loc) · 1.4 KB

dragnn_ops.bulk_fixed_feature_ids(handle, num_channels, component=None, num_ids_per_channel=None, name=None)

dragnn_ops.bulk_fixed_feature_ids(handle, num_channels, component=None, num_ids_per_channel=None, name=None)

Defined in tensorflow/dragnn/core/ops/gen_dragnn_bulk_ops.py.

This op is a variant of BulkFixedFeatures that only outputs fixed feature IDs.

Each fixed feature channel must produce exactly |num_ids_per_channel| feature IDs per step, and feature weights are ignored.

Args:

  • handle: A Tensor of type string. handle to a ComputeSession.
  • num_channels: An int that is >= 1. number of feature channels in the component.
  • component: An optional string. Defaults to "". name of the component to run.
  • num_ids_per_channel: An optional int. Defaults to 1. number of feature IDs extracted per channel at each step.
  • name: A name for the operation (optional).

Returns:

A tuple of Tensor objects (output_handle, ids). * output_handle: A Tensor of type string. handle to updated ComputeSession. * ids: A list of num_channels Tensor objects of type int64. a list of num_channels [batch_size, num_steps, num_ids_per_channel] tensors of feature IDs, one per channel. The feature IDs for each batch item are padded with trailing -1s to the maximum number of steps across the batch.