This paper explores the effectiveness of Supervised Fine-Tuning (SFT) as a method for knowledge injection into Large Language Models (LLMs), specifically focusing on recent sporting events. The study compares two dataset generation strategies—token-based and fact-based scaling—to create training data that helps the model learn new information. Experiments on GPT-4 demonstrate that while token-based scaling can improve Q&A accuracy, it may not provide uniform coverage of new knowledge. In contrast, fact-based scaling offers a more systematic approach to ensure even coverage across all facts. The paper presents a novel dataset generation process that leads to more effective knowledge ingestion through SFT, showing significant performance improvements in Q&A tasks related to out-of-domain knowledge. The study contributes to the understanding of domain adaptation for LLMs and highlights the potential of SFT in enhancing the factuality of LLM responses in specific knowledge domains.This paper explores the effectiveness of Supervised Fine-Tuning (SFT) as a method for knowledge injection into Large Language Models (LLMs), specifically focusing on recent sporting events. The study compares two dataset generation strategies—token-based and fact-based scaling—to create training data that helps the model learn new information. Experiments on GPT-4 demonstrate that while token-based scaling can improve Q&A accuracy, it may not provide uniform coverage of new knowledge. In contrast, fact-based scaling offers a more systematic approach to ensure even coverage across all facts. The paper presents a novel dataset generation process that leads to more effective knowledge ingestion through SFT, showing significant performance improvements in Q&A tasks related to out-of-domain knowledge. The study contributes to the understanding of domain adaptation for LLMs and highlights the potential of SFT in enhancing the factuality of LLM responses in specific knowledge domains.