Hardcoded pyramid_levels causing runtime error with stride customization

[hanoglu]

There is no option in the training code to change the stride variable. Apparently, stride depends on pyramid_levels value. But this value is
hardcoded in the line
self.anchor_points = AnchorPoints(pyramid_levels=[3,], row=row, line=line)
and in case it is changed, it gives this runtime error;
RuntimeError: The size of tensor a (1024) must match the size of tensor b (256) at non-singleton dimension
at
output_coord = regression + anchor_points