LIBSVM easy.py valueerror:need more than 0 values to unpack Windows Ultimate solution

The phenomenon is:

Python easy.py train test

Output:

Scaling Training Data ...
Warning:original #nonzeros 100389
New #nonzeros 9983948
Use-l 0 If many original feature values are zeros
Cross validation ...
Traceback (most recent):
File "easy.py", line at <module>
C,g,rate = Map (Float,last_line.split ())
Valueerror:need more than 0 values to unpack

grid.py, path and so on are no problem.

The final reason is located in:

cmd = ' python {0}-svmtrain ' {1} '-gnuplot ' {2} ' ' {3} ' '. Format (grid_py, Svmtrain_exe, Gnuplot_exe, Scaled_file)
The reason is that grid.py is opened as a text file, and the Python character can be considered a python file to run.

Via http://bbs.pinggu.org/forum.php?mod=viewthread&action=printable&tid=1529067

LIBSVM easy.py valueerror:need more than 0 values to unpack Windows Ultimate solution