code for generating training data used for fine-tuning the LLM - View it on GitHub