Reviewers find the ideas of exploration to find high disagreement between the dynamic models interesting and appreciate the sample complexity analysis, albeit the proof follows prior work closely. The empirical evaluation is strong enough to support the proposed algorithm in practical environments.