[QA] Beyond Model Collapse: Scaling Up with Synthesized Data Requires Reinforcement

Arxiv Papers - [QA] Beyond Model Collapse: Scaling Up with Synthesized Data Requires Reinforcement