The paper provides a nice adaptation of deep kernel learning to the few-shot setting, with promising performance over key deep learning baselines. Reviewers are united in their support for the work. Please carefully consider reviewer comments (and post rebuttal updates) in preparing final revisions.