Is One Layer Enough? Training A Single Transformer Layer Can Match Full-Parameter RL Training Paper • 2607.01232 • Published 28 days ago • 8