CODEWITHSUNDEEP
pythontensorflow
Mj1992
Mj1992

Reputation: 3504

Tensorflow concat 2D dimension Tensor to a 3D dimension tensor

This is rather a simple question. I've a tensor in 3D shape. Lets say it is [2,2,3]. I've another tensor in 2D shape [2,2]. But the first 2 dimensions of both the elements are matching. I want to concat them such that the 2D tensor is added to the 3D tensor in the third dimension. i.e [2,2,4].

I am not sure how to achieve this. I tried using tf.concat and tf.stack but that requires both the tensors to be the same rank.

Upvotes: 2

Views: 3314

Answers (1)

mrry
mrry

Reputation: 126154

You can use tf.expand_dims() to convert the smaller tensor to the correct rank, followed by tf.concat():

tensor_3d = tf.placeholder(tf.float32, shape=[2, 2, 3])
tensor_2d = tf.placeholder(tf.float32, shape=[2, 2])

tensor_2d_as_3d = tf.expand_dims(tensor_2d, 2)  # shape: [2, 2, 1]

result = tf.concat([tensor_3d, tensor_2d_as_3d], 2)  # shape: [2, 2, 4]

Upvotes: 7

Related Questions