Harry Harrison

User questions LLM prose quality after four years of development

A user on the r/LocalLLaMA subreddit is questioning the ability of current large language models (LLMs) to produce high-quality prose, even after extensive fine-tuning. Despite experimenting with various datasets and tr…