Testing error

[Estelle Quan]

Hi,
I have some problems on testing
I have successfully train a simple digit classifier by http://eblearn.cs.nyu.edu:21991/doku.php?id=beginner_tutorial2_train which didn't describe how to test on the classer.
I imitate style of writing by the bottom of http://eblearn.cs.nyu.edu:21991/doku.php?id=svhn_tutorial,
but it run it ( as attachments )had mistake about :

arch: loaded 22 modules.

Random seed initialized to 0
Concatenated 1 matrices into 1: 19415 from [ /home/qlj/Public/eblearn/dssp/net00050.mat ]
Loaded weights from 19415: (x:[ 19415 ],dx:[ 19415 ],ddx:[ 19415 ])
forward weights: (x:(19415 min -1.00658 max 1.23602),dx: (19415 min 0 max 0),ddx: (19415 min 0 max 0))
Targets: 3x3
[[ 1 -1 -1 ] [ -1 1 -1 ] [ -1 -1 1 ]]
Training with: trainer module : energy l2_energy is the euclidean distance between inputs, class_answer module class_answer with 3 classes, confidence type 2 and targets 3x3.
Setting progress file to "progress"
Gradient parameters: eta 0.0001 stopping threshold 0 decay_l1 0 decay_l2 0 decay_time 0 inertia 0 anneal_value 0 annueal_period 0 gradient threshold 0
Test only mode...
Segmentation fault (core dumped)

Would you please help me to check this mistake, thank you very much!