The Beyond the Imitation Game Benchmark (BIG-bench) is a collaborative benchmark intended to probe large language models and extrapolate their future capabilities. Big-bench include more than 200 tasks. Image source: https://arxiv.org/pdf/2206.04615.pdf