as `tf.tpu.experimental.embedding.serving_embedding_lookup` which can
        take arbitrary rank of dense and sparse tensor. For ragged tensor,
        though the input tensor remains to be rank 2, the activations now can be
        rank 2 or above by specifying the output shape in the feature config or
        via the build method.

*   Add

```python
    # Scalar indices (output is rank(params) - 1).
    output[a_0, ..., a_n, b_0, ..., b_n] =
      params[a_0, ..., a_n, indices, b_0, ..., b_n]

    # Vector indices (output is rank(params)).
    output[a_0, ..., a_n, i, b_0, ..., b_n] =
      params[a_0, ..., a_n, indices[i], b_0, ..., b_n]

    # Higher rank indices (output is rank(params) + rank(indices) - 1).