Bootstrapping Post-training Signals for Open-ended Tasks via Rubric-based Self-play on Pre-training Text

arXiv:2604.20051v1 Announce Type: cross Abstract: Self-play has recently emerged as a promising paradigm to train Large Language Models (LLMs). In self-play, the target LLM creates the task input (e.g., ask a question), which it then addresses itself by producing a task output (e.g., give an answer)