Hacker News new | ask | show | jobs
by stevekemp 3112 days ago
Reading source code of existing learning/real chains. Then trial and error - the usual :)