BIG-bench

Collaborative benchmark for measuring and extrapolating language model capabilities.