datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
TwoHopFactThis is the dataset introduced in the paper Do Large Language Models Latently Perform Multi-Hop Reasoning?.
Code: https://github.com/google-deepmind/latent-multi-hop-reasoning.
SOCRATESThis is the dataset introduced in the paper Do Large Language Models Latently Perform Multi-Hop Reasoning without Exploiting Shortcuts?.
Code: https://github.com/google-deepmind/latent-multi-hop-reasoning.