Error using svmpredict on libsvm

Hi all,

Once I want to predict my following vectors of values in order to get the correct label, I get the following message.

 "Length of label vector does not match # of instances.\n"

I have a multiclass SVM where labels goes from 1 to 9 so 9 rows of one column (9x1). I would like to examine this vector of labels with a single vector of row and n features. Is it possible? Why should I need the same number of rows from labels and vectors of testing? Can I modify the svmpredict.c file?

Thanks in advanced.

Raúl.