Comment by chromoblob

what about directly connecting two "powerful" human learning models directly? can you do this without disrupting your determinacy dynamics?