I found that there are several projects about ffm in your github, such as libffm-ftrl,Difacto_DMLC.Which one would you recommend to use? Do you have any data about their effectiveness and efficiency on benchmark dataset?
libffm-ftrl is a single machine version with ftrl updater to produce sparse features
however, Difacto_DMLC is a distributed version with Parameter Server of fm not ffm
I found that there are several projects about ffm in your github, such as libffm-ftrl,Difacto_DMLC.Which one would you recommend to use? Do you have any data about their effectiveness and efficiency on benchmark dataset?