Training Regime Day 0: Introduction — LessWrong