Avatarl: Training langauge models from scratch with pure RL

2 points | by krkartikay 4 months ago