Number of training examples:   10000
Number of validation examples:  5000 (for model selection)
Number of test examples:       10000

1. Reported is the test error (TeVErr), for the model selected such
   that the validation error (ValErr) is smallest (independent on each of
   the five realizations of the data set.).

run:1  ValErr:0.0334  d1:3  d2:1  C:0.80  l:3  =>  TeVErr:0.0362  svs=1873
run:2  ValErr:0.0334  d1:2  d2:2  C:3.00  l:3  =>  TeVErr:0.0389  svs=3224
run:3  ValErr:0.0352  d1:2  d2:2  C:4.00  l:5  =>  TeVErr:0.0390  svs=3327
run:4  ValErr:0.0356  d1:2  d2:2  C:10.0  l:4  =>  TeVErr:0.0378  svs=3228
run:5  ValErr:0.0380  d1:2  d2:2  C:2.00  l:3  =>  TeVErr:0.0406  svs=3099

   TeVErr average+/-std. deviation: 3.85 +/- 0.16 %

2. Reported is the test error (TeVErr), for the model selected such
   that the validation error (ValErr) is smallest. ValErr, TeVErr and
   #svs are averaged over the five runs.

ValErr:0.0360  d1:2  d2:2  C:2.00  l:3  =>  TeVErr:0.0383  #svs=3058