I'm not very confident, but I think w and V are linear and pairwise coefficients, respectively, and so, l2_reg_w and l2_reg_V should be penalty weights for linear and pairwise coefficients.
Currently, the explanations of them seem wrongly swapped. This PR proposes to fix them.
I'm not very confident, but I think
w
andV
are linear and pairwise coefficients, respectively, and so,l2_reg_w
andl2_reg_V
should be penalty weights for linear and pairwise coefficients.Currently, the explanations of them seem wrongly swapped. This PR proposes to fix them.