CODEWITHSUNDEEP
pythonnumpytensorflowdataset
Dharanikumar Sekar
Dharanikumar Sekar

Reputation: 21

How to convert list of tuple (tf.constant, tf.constant) as tensorflow dataset?

I'm trying to create a Transformer Model, I have 2 np.arrays, both have strings, I used them to create a list of tuples

The format of the tuple is : 

class 'tuple' (tf.Tensor: shape=(), dtype=string, numpy=b'abc', tf.Tensor: shape=(), dtype=string, numpy=b'xyz')

I want to combine these tuples to form tensorflow.python.data.ops.dataset_ops._OptionsDataset, how can I do that?

Or is there any other way I could do it?

New to this, thanks for your help!

Upvotes: 1

Views: 1581

Answers (1)

Rishabh Sahrawat
Rishabh Sahrawat

Reputation: 2507

You can use tensor_from_slices. Here.

EDIT

Example
# Two tensors can be combined into one Dataset object. 
values1 = tf.constant(['A', 'B', 'A']) # ==> 3x1 tensor
values2 = tf.constant(['A', 'B', 'A']) # ==> 3x1 tensor
dataset = Dataset.from_tensor_slices((values1, values2))

Upvotes: 2

Related Questions