[2607.01232] Is One Layer Enough? Training A Single Transformer Layer Can Match Full-Parameter RL Training

X Facebook はてブ LINE Pinterest コピー