【论文阅读】AdaCtrl: Towards Adaptive and Controllable Reasoning via Difficulty-Aware Budgeting
TowardsAdaptiveandControllableReasoningviaDifficulty-AwareBudgeting3Method3.1长度触发标签作为控制接口(Length-TriggerTagsasControllingInterface)3.2冷启动微调(Cold-startfine-tuning)3.3难度感知的强化学习框架