how do i find the error rates at different iterations for logregfit?

the overall error rate can be found is by :=

model = logregFit(X, y, 'lambda', 1, 'preproc', pp);

[yhat, p] = logregPredict(model, X);

errRate(1) = mean(yhat ~= y);

where X is training set and y is labels, but how to find error rate at different iterations(for, eg say at iteration 100 the error rate would be ...