Fine-tune small foundational LLM on typical large enterprise use-case and compare results with pre-trained and large scale models. Instruction dataset will be generated artificially with a SOTA LLM.