Training, testing and validation datasets in Matlab Neural Networks

I am trying to setup a neural network, it is supposed to have 3 inputs vectors, each one with 10 inputs, that should be read as described in the manual in batch mode in opposition to the sequential input for dynamic networks.

I have managed to setup everything correctly, but the training method is not being able to divide the inputs correctly into training, testing and validation datasets, from what i understand in the "tr" variable resulting from the training procedure, all input vectors are assigned to the train dataset.

Does anyone have an idea on how to set this up?

Example data:

inputs = 
 
 [10x1 double]    [10x1 double]    [10x1 double]
 [10x1 double]    [10x1 double]    [10x1 double]
 [10x1 double]    [10x1 double]    [10x1 double]
 

where

cell2mat(inputs(1,1))
 
 -12.4523
   0.2072
  13.4163
   7.3567
  -0.1878
  -2.8456
 -10.2632
  10.0383
   9.5882
  -0.6623
 

and

 targets = { 0 1 0 }
 

Thanks in advance,

SwatchPuppy

---