Do well-written, clear instructions beat few-shotting for tiny-LLMs? | Dark Hacker News