FINE-TUNING LLMS IS A HUGE WASTE OF TIME

Summary from the TLDR Web Dev newsletter

Fine-tuning advanced LLMs for knowledge injection is a waste of time and potentially destructive. Updating neurons in these models can overwrite existing valuable knowledge, leading to unintended consequences.