Chain-of-Thought Hub: A Continuous Effort to Measure Large Language Models' Reasoning Performance | Synapse