This paper proposes a new adaptive sampling and clustering framework for learning stochastic block models for network data. The paper provides a comprehensive theoretical analysis of the framework, as well as empirical results, that clearly demonstrate the useful contributions of the approach. Reviewers were in agreement that the paper is worthy of acceptance for the conference. There were some comments from reviewers about opportunities to improve the presentation in the paper. The authors are strongly encouraged to take these reviewer comments into account when revising the paper for the final version.