· ai
Beyond the Imitation Game: Quantifying and extrapolating the capabilities oflanguage models
Summary - Researchers assembled BIG-bench, a collection of 204 tasks created by many contributors to evaluate current and future language model capabilities. -...