This repository contains the dataset described in "Morables : A Benchmark for Assessing Abstract Moral Reasoning in LLMs with Fables", which is due to be presented at EMNLP 2025.
Each fable has an associated free-text moral, sourced from various websites and books (detailed in the paper). It is intended for use in NLP text understanding and moral inference tasks.
