Hacker News Reader

Which dataset formats are you planning to support? (OpenAI, ShareGPT, Alpaca, Context-based ...)

Same as OpenAI, where it's either a json or jsonl file with {"input": "text", "output": "text"}