This is a collection designed to present the ten RLLM steps/ training runs intended to improve Phi-1.5's outputs towards coherence and politeness.