This is the Mystery Zebra dataset created as part of the paper "Lexical Recall or Logical Reasoning: Probing the Limits of Reasoning Abilities in Large Language Models". We make the dataset available in a .csv format for your convenience. The code used to generate the puzzles in this dataset can be found in:
https://github.com/arg-tech/MysteryZebra The structure of the dataset is straightforward and easy to parse. In the following, we detail the content of all… See the full description on the dataset page:
https://huggingface.co/datasets/arg-tech/MysteryZebra.
