SSealbook

google/BIG-bench

Beyond the Imitation Game collaborative benchmark for measuring and extrapolating the capabilities of language models