ActuarialMathBench is a domain-specific benchmark dataset designed to evaluate the mathematical reasoning capabilities of Large Language Models (LLMs) within the actuarial domain. The dataset consists of 750 question-answer pairs derived from Society of Actuaries (SOA) sample exams.
Each exam subject is part of the curriculum to obtain the designation Associate of the Society of Actuaries (ASA). The % Pass Mark is the 90th percentile pass mark percentage over the… See the full description on the dataset page:
https://huggingface.co/datasets/BjornvanBraak/ActuarialMathBench.
