ktop | visualized monitoring dashboard | Monitoring library

I have a problem with using Tensorflow. I have four images with their corresponding indices. I want to make an image from them. I tried for loops, tf.gather, tf.assign, and so on but all show error. If somebody help me, it would be really appreciated. I explain my question with one small example: We have 4 tensors and their indices from tensor tf.ktop function: (I write like MATLAB for just for simplicity)

a = [1, 2; 5, 6] a_idx = [0, 1; 2, 3] b = [3, 4; 7, 8] b_idx = [0, 1; 2, 3] c = [9, 10; 13, 14] c_idx = [0, 1; 2, 3] d = [11, 12; 15, 16] d_idx = [0, 1; 2, 3]

I am looking for a big image from a, b, c, and d and their indices like:

image = [a b; c d] image = [1, 2, 3, 4; 5, 6, 7, 8;9 10, 11, 12;13, 14, 15, 16]

In python I have something like:

I want to show an image in the tablet of Pepper from his camera. I have done the Javascript code to process and show the image in a web view. The problem is that the image showed is strange:

I have written this code based on this example:

Recently, I start learning torch. But, the following question really stumps me. A seq2seq demo is available here, and the definition of model is showed as follow: