Question about libsvm training output

I've been using libsvm to train support vector machines, and when I train it, it produces output like:

........* optimization finished, #iter = 8594 obj = -14403.958604, rho = 2.269881 nSV = 6262, nBSV = 6195 Total nSV = 6262 

Can I infer anything useful from the obj and rho values? For example, if they get closer to zero, does that mean the model is fitting the training data very well or any such thing? I usually score the model against some separate test data to measure it, but I figured it must output these values because they are important, yet their meaning escapes me.