Fetching the paper…

Do Large Language Models Perform Latent Multi-Hop Reasoning without Exploiting Shortcuts? · Around