Activation Control for Efficiently Eliciting Long Chain-of-thought Ability of Language Models

Open in new window