libsvm python, svmulti_predict error

Question

[honglienqn]

I am using libsvm python version 3-18. When I call the function svm_predict in module 'svmutil', I have the error:

Traceback                
 __call__    C:\Program Files (x86)\PyScripter\Lib\rpyc-python2x.zip\rpyc\core\netref.py    123        
  syncreq    C:\Program Files (x86)\PyScripter\Lib\rpyc-python2x.zip\rpyc\core\netref.py    45        
  sync_request    C:\Program Files (x86)\PyScripter\Lib\rpyc-python2x.zip\rpyc\core\protocol.py      343        
  serve    C:\Program Files (x86)\PyScripter\Lib\rpyc-python2x.zip\rpyc\core\protocol.py    305        
  _recv    C:\Program Files (x86)\PyScripter\Lib\rpyc-python2x.zip\rpyc\core\protocol.py    265        
   recv    C:\Program Files (x86)\PyScripter\Lib\rpyc-python2x.zip\rpyc\core\channel.py    36        
   read    C:\Program Files (x86)\PyScripter\Lib\rpyc-python2x.zip\rpyc\core\stream.py    105        
exceptions.EOFError: [Errno 10054] An existing connection was forcibly closed by the remote host

This is my full call functions:

y, x = svm_read_problem('D:\libsvm-3.18\Test_examples\output_train.txt')
m1 = svm_train (y, x, '-s 1 -t 2 -c 0.03125 -g 0.0078125')
svm_save_model('D:\libsvm-3.18\Test_examples\m1.model', m1)
p_labels, p_acc, p_vals = svm_predict(y, x, m1)

I don't know what causes this error and how to fix this.Thanks for your help and consideration to fix this error.

Regards,
lotus