arrays - Binary mask in Tensorflow -

i mask every other value along particular dimension of tensor don't see way generate such mask. example

#masking on 2nd dimension = [[1,2,3,4,5],[6,7,8,9,0] mask = [[1,0,1,0,1],[1,1,1,1,1]] b = * mask #would return [[1,0,3,0,5],[6,0,8,0,0]] 

is there easy way generate such mask?

ideally following:

mask = tf.ones_like(input_tensor) mask[:,::2] = 0 mask * input_tensor 

but slice assigning doesn't seem simple in numpy.

you can programmatically create such tensor mask using python. convert tensor. there's no such support in tensorflow api. tf.tile([1,0], num_of_repeats) might fast way create such mask not great either if have odd number of columns.

(btw, if end creating boolean mask, use tf.boolean_mask())