ml has existed for like 50 or so years, since basically first computers. if your model does not have billions of parameters (in this case, for 2 species, only 9 such parameters were identified), it uses far less compute.