An op enabling differentiation of TPU Embeddings.
This op simply returns its first input, which is assumed to have been sliced from the Tensors returned by TPUEmbeddingDequeueActivations. The presence of this op, and its first argument being a trainable Variable, enables automatic differentiation of graphs containing embeddings via the TPU Embedding Python libraries.
Constants
String | OP_NAME | The name of this op, as known by TensorFlow core engine |
Public Methods
Output<TFloat32> |
asOutput()
Returns the symbolic handle of the tensor.
|
static TPUEmbeddingActivations | |
Output<TFloat32> |
output()
|
Inherited Methods
Constants
public static final String OP_NAME
The name of this op, as known by TensorFlow core engine
Public Methods
public Output<TFloat32> asOutput ()
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.
public static TPUEmbeddingActivations create (Scope scope, Operand<TFloat32> embeddingVariable, Operand<TFloat32> slicedActivations, Long tableId, Long lookupId)
Factory method to create a class wrapping a new TPUEmbeddingActivations operation.
Parameters
scope | current scope |
---|---|
embeddingVariable | A trainable variable, enabling optimizers to find this op. |
slicedActivations | The embedding activations Tensor to return. |
tableId | The id of the table in the embedding layer configuration from which these activations were computed. |
lookupId | Identifier of the set of embedding indices which produced these activations. |
Returns
- a new instance of TPUEmbeddingActivations