Reinforcement Learning via Self-Distillation 🎥 📝 ICLR Workshop on Scaling Post-training for LLMs, Rio de Janeiro