tf.raw_ops.LoadTPUEmbeddingFTRLParametersGradAccumDebug

Load FTRL embedding parameters with debug support.

tf.raw_ops.LoadTPUEmbeddingFTRLParametersGradAccumDebug(
    parameters, accumulators, linears, gradient_accumulators, num_shards, shard_id,
    table_id=-1, table_name='', config='', name=None
)

An op that loads optimization parameters into HBM for embedding. Must be preceded by a ConfigureTPUEmbeddingHost op that sets up the correct embedding table configuration. For example, this op is used to install parameters that are loaded from a checkpoint before a training loop is executed.

Args:

  • parameters: A Tensor of type float32. Value of parameters used in the FTRL optimization algorithm.
  • accumulators: A Tensor of type float32. Value of accumulators used in the FTRL optimization algorithm.
  • linears: A Tensor of type float32. Value of linears used in the FTRL optimization algorithm.
  • gradient_accumulators: A Tensor of type float32. Value of gradient_accumulators used in the FTRL optimization algorithm.
  • num_shards: An int.
  • shard_id: An int.
  • table_id: An optional int that is >= -1. Defaults to -1.
  • table_name: An optional string. Defaults to "".
  • config: An optional string. Defaults to "".
  • name: A name for the operation (optional).

Returns:

The created Operation.