tf.raw_ops.RecvTPUEmbeddingActivations

An op that receives embedding activations on the TPU.

tf.raw_ops.RecvTPUEmbeddingActivations(
    num_outputs, config, name=None
)

The TPU system performs the embedding lookups and aggregations specified by the arguments to TPUEmbeddingEnqueue(Integer/Sparse/SparseTensor)Batch. The results of these aggregations are visible to the Tensorflow Graph as the outputs of a RecvTPUEmbeddingActivations op. This op returns a list containing one Tensor of activations per table specified in the model. There can be at most one RecvTPUEmbeddingActivations op in the TPU graph.

Args:

  • num_outputs: An int that is >= 1. The number of output activation tensors, equal to the number of embedding tables in the model.
  • config: A string. Serialized TPUEmbeddingConfiguration proto.
  • name: A name for the operation (optional).

Returns:

A list of num_outputs Tensor objects with type float32.