Class CollateTPUEmbeddingMemory

java.lang.Object
org.tensorflow.op.RawOp
org.tensorflow.op.tpu.CollateTPUEmbeddingMemory
All Implemented Interfaces:
Shaped, Op, Operand<TString>

@Operator(group="tpu") public final class CollateTPUEmbeddingMemory extends RawOp implements Operand<TString>
An op that merges the string-encoded memory config protos from all hosts.
  • Field Details

  • Constructor Details

    • CollateTPUEmbeddingMemory

      public CollateTPUEmbeddingMemory(Operation operation)
  • Method Details

    • create

      @Endpoint(describeByClass=true) public static CollateTPUEmbeddingMemory create(Scope scope, Iterable<Operand<TString>> memoryConfigs)
      Factory method to create a class wrapping a new CollateTPUEmbeddingMemory operation.
      Parameters:
      scope - current scope
      memoryConfigs - String-encoded memory config protos containing metadata about the memory allocations reserved for TPUEmbedding across all hosts.
      Returns:
      a new instance of CollateTPUEmbeddingMemory
    • mergedMemoryConfig

      public Output<TString> mergedMemoryConfig()
      Gets mergedMemoryConfig.
      Returns:
      mergedMemoryConfig.
    • asOutput

      public Output<TString> asOutput()
      Description copied from interface: Operand
      Returns the symbolic handle of the tensor.

      Inputs to TensorFlow operations are outputs of another TensorFlow operation. This method is used to obtain a symbolic handle that represents the computation of the input.

      Specified by:
      asOutput in interface Operand<TString>
      See Also: