Foundation Models, RLAIF, SFT, pre-training, model architecture, LLM optimisation, distributed training, LLM training, large-scale GPU clusters, RLHF