Evaluate feature representations and transfer learning model architectures on a wide variety of target tasks.

benchmarking data-science machine-learning