Reinforcing Recursive Language Models

RL fine-tuning small models to behave as recursive language models.