The comparison rates TensorFlow higher than Torch on two criteria and lower than Torch on two, and explicitly agrees with you about Lua ("However, let's face it

The comparison rates TensorFlow higher than Torch on two criteria and lower than Torch on two, and explicitly agrees with you about Lua ("However, let's face it, Lua is not yet a mainstream language."). What, in your view, should it have looked like in order not to be a "Torch puff piece"? [EDITED to fix a trivial typo.]