aitraining method

1 triples
GPTKB property

Random triples
Subject Object
gptkb:Open_AI_Dota_2 reinforcement learning with human feedback