The Truth is in There: Improving Reasoning in Language Models with Layer-Selective Rank Reduction

Open in new window