Avatarl: Training language models from scratch with pure reinforcement learning

2 points | by neehao 3 days ago