tf.raw_ops.XlaSparseDenseMatmulGradWithAdamAndCsrInput 

tf.raw_ops.XlaSparseDenseMatmulGradWithAdamAndCsrInput(
    row_pointers,
    sorted_sample_ids,
    sorted_token_ids,
    sorted_gains,
    activation_gradients,
    learning_rate,
    embedding_table,
    momenta,
    velocity,
    num_minibatches_per_physical_sparse_core,
    use_sum_inside_sqrt,
    beta1,
    beta2,
    epsilon,
    table_name,
    clip_weight_min=float('-inf'),
    clip_weight_max=float('inf'),
    name=None
)

row_pointers A Tensor of type int32.
sorted_sample_ids A Tensor of type int32.
sorted_token_ids A Tensor of type int32.
sorted_gains A Tensor of type float32.
activation_gradients A Tensor of type float32.
learning_rate A Tensor of type float32.
embedding_table A Tensor of type float32.
momenta A Tensor of type float32.
velocity A Tensor of type float32.
num_minibatches_per_physical_sparse_core A Tensor of type int32.
use_sum_inside_sqrt A bool.
beta1 A float.
beta2 A float.
epsilon A float.
table_name A string.
clip_weight_min An optional float. Defaults to float('-inf').
clip_weight_max An optional float. Defaults to float('inf').
name A name for the operation (optional).

A tuple of Tensor objects (updated_embedding_table, updated_momenta, updated_velocity).
updated_embedding_table A Tensor of type float32.
updated_momenta A Tensor of type float32.
updated_velocity A Tensor of type float32.