ValueError: sample larger than population

[SeekPoint]

mldl@mldlUB1604:/ub16_prj/NTM-One-Shot-TF$ python Omniglot.py
Shapes Recieved in Update: V, dim, val are ==> [64, 128] [64] [64]
Shapes Recieved in Body of Update: v, d2, chg are ==> [128] [] []
wlu_tm1 : [16, 4]
Shapes Recieved in Update: V, dim, val are ==> [16, 128, 40] [16] [16, 40]
Shapes Recieved in Body of Update: v, d2, chg are ==> [128, 40] [] [40]
Compiling the Model
Output, Target shapes: [16, 50, 5] [16, 50, 5]
Done
Training the model
Traceback (most recent call last):
File "Omniglot.py", line 112, in
omniglot()
File "Omniglot.py", line 85, in omniglot
for i, (batch_input, batch_output) in generator:
File "/home/mldl/ub16_prj/NTM-One-Shot-TF/MANN/Utils/Generator.py", line 36, in next
return (self.num_iter - 1), self.sample(self.nb_samples)
File "/home/mldl/ub16_prj/NTM-One-Shot-TF/MANN/Utils/Generator.py", line 41, in sample
sampled_character_folders = random.sample(self.character_folders, nb_samples)
File "/usr/lib/python2.7/random.py", line 323, in sample
raise ValueError("sample larger than population")
ValueError: sample larger than population
mldl@mldlUB1604:/ub16_prj/NTM-One-Shot-TF$ ll data/
total 18412
drwxrwxr-x 5 mldl mldl 4096 8月 4 18:24 ./
drwxrwxr-x 7 mldl mldl 4096 8月 4 18:16 ../
drwxr-xr-x 32 mldl mldl 4096 10月 21 2015 images_background/
-rw-rw-r-- 1 mldl mldl 1319022 8月 4 18:23 images_background_small1.zip
-rw-rw-r-- 1 mldl mldl 1580917 8月 4 18:23 images_background_small2.zip
-rw-rw-r-- 1 mldl mldl 9464212 8月 4 18:23 images_background.zip
-rw-rw-r-- 1 mldl mldl 6462886 8月 4 18:23 images_evaluation.zip
drwxrwxr-x 2 mldl mldl 4096 8月 3 17:26 omniglot/
drwxrwxr-x 2 mldl mldl 4096 8月 4 18:23 one-shot-classification/
mldl@mldlUB1604:~/ub16_prj/NTM-One-Shot-TF$ ll data/images_background