https://graphlogic.ai/blog/ai-chatbots/ai-fundamentals/arc-benchmark-evaluating-llms-reasoning/