KUM-Bench (Kyoto University Math Entrance Exam Benchmark) is designed to evaluate advanced Japanese reasoning capabilities by leveraging mathematics entrance exam questions from Kyoto University—one of the most prestigious universities in Japan. These problems require a high level of problem-solving skills and serve as an excellent resource for testing Large Language Models (LLMs) in Japanese reasoning tasks.… See the full description on the dataset page:
https://huggingface.co/datasets/Inoichan/KUM-Bench.
