python wrapper - Cost for each sample

[skeld]

Hello,
I'm working with the python wrapper bhtsne.py, the iris dataset, and on a ubuntu 14.04.
I'm trying to get the costs for each samples (as specified iat the end of bh_tsne()).

I de-commented the last line (of the function)

#read_unpack('{}d'.format(sample_count), output_file)

and adapted it

_read_unpack('{}d'.format(len(results)), output_file)

and put it before the yield, inside a simple print()

However, all the cost are equal to zero. Even when i set a very low number of iterations, and the verbose tell me that the error is still high:

$ ./bhtsne.py -d 2 -p 30 -v -i iris_data.txt -o tsne_test.output --no_pca -m 100
Read the 150 x 4 data matrix successfully!
Using current time as random seed...
Using no_dims = 2, perplexity = 30.000000, and theta = 0.500000
Computing input similarities...
Building tree...
 - point 0 of 150
Input similarities computed in 0.01 seconds (sparsity = 0.706622)!
Learning embedding...
Iteration 50: error is 45.556438 (50 iterations in 0.02 seconds)
Iteration 99: error is 44.807590 (50 iterations in 0.02 seconds)
Fitting performed in 0.04 seconds.
Wrote the 150 x 2 data matrix successfully!
('costs: ', (0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, [... all 0.0 ..], 0.0, 0.0))

What did i missed' or how can i get the costs ?

Thanks