"Spittin' Bars: Fine-tuning Large Language Models for Lyric Generation" demonstrates notable differences in text generation between pre and post fine-tuned large language models. The fine-tuned model consistently produced text aligning with the stylistic and thematic patterns found in song lyrics. Utilizing a proprietary dataset created by scraping lyrics from thirty songs spanning diverse genres—including classic rock, reggae, drill, psychedelic, indie, country, emo rap, soul, pop, electronic, bluegrass, R&B, new wave, and hip hop—the model's fine-tuning process demonstrates significant impact on its ability to generate text resembling song lyrics.