Is One Layer Enough? A Single Transformer Layer Matches Full-Parameter RL Train

Abstract page for arXiv paper 2607.01232: Is One Layer Enough? Training A Single Transformer Layer Can Match Full-Parameter RL Training