Chapter 13: How to convert a sparse tensor with rank(st_input) greater than 2 to RaggedTensor?

I am reading a text file through Data API and converting the text into TFRecords. When I am reading the records line by line, I am getting a sparse matrix of rank 2 which is easily being transformed into a RaggedTensor by `tf.RaggedTensor.from_sparse`. However, when I am reading the records in batch of size 10, resulting sparse tensor is of rank 3. In the documentation, it is written that `tf.RaggedTensor.from_sparse` works only with sparse tensors of rank 2. Is there a way to create RaggedTensors from a sparse tensor of rank 3? I tried `tf.sparse.reshape` and tried removing the axis=1. This was successful as I was able to print the output. But this seems a little hackish way to do and I am looking for something more tensorflow-ish. An example to reproduce my case is as follows:-

1. When reading tfrecords line by line-

```python
indices = array([
[ 0, 0],
[ 0, 1],
[ 0, 2],
[ 0, 3],
[ 0, 4],
[ 0, 5],
[ 0, 6],
[ 0, 7],
[ 0, 8],
[ 0, 9],
[ 0, 10]], dtype=int64)

values = array([b'What', b'interesting', b'fact', b'about', b'India', b'can',
b'you', b'add', b'to', b'my', b'knowledge?'], dtype=object)
dense_shape = array([ 1, 11], dtype=int64)

tf.RaggedTensor.from_sparse(tf.SparseTensor(indices=indices,values=values,dense_shape=dense_shape))
```

gives

```

```

2. When reading a batch of size 1 (actual batch size was 10 but that would become way too cumbersome):

```python
indices = array([
[ 0, 0, 0],
[ 0, 0, 1],
[ 0, 0, 2],
[ 0, 0, 3],
[ 0, 0, 4],
[ 0, 0, 5],
[ 0, 0, 6],
[ 0, 0, 7],
[ 0, 0, 8],
[ 0, 0, 9],
[ 0, 0, 10]], dtype=int64)

values = array([b'What', b'interesting', b'fact', b'about', b'India', b'can',
b'you', b'add', b'to', b'my', b'knowledge?'], dtype=object)
dense_shape = array([ 1, 1, 11], dtype=int64)

tf.RaggedTensor.from_sparse(tf.SparseTensor(indices=indices,values=values,dense_shape=dense_shape))
```

gives an error message:

```
ValueError: rank(st_input) must be 2
```

However, using the statement

```python
tf.sparse.reshape(tf.SparseTensor(indices=indices,values=values,dense_shape=dense_shape),shape=[1,11])
```

resolves this issue.

Apologies for the sloppy editing.