Fetching the paper…

Self-Play Fine-Tuning Converts Weak Language Models to Strong Language Models · Around