LIBSVM vs sklearn

I’m trying to port the SVC-based modelling from python/sklearn to Julia using the LIBSVM.jl package. However, I noticed one strange thing. Sklearn SVC outperforms LIBSVM.jl, although both are based on the same libsvm library. Also, when I try to calculate probabilities after training models, I get different values for LIBSVM and sklearn with the same model parameters. Has anyone encountered the same behavior? How can this difference be explained?