R1: Comparison with inexact methods Aligning with prior exact papers [10, 18], we focus on comparisons with exact

We thank all five reviewers for their detailed and incisive feedback. We tested AustereMH [16], an inexact method, on robust linear regression in Section 5.1 with N = 5000. We added this to the Appendix. R2: What if E[B] > N E[B] is typically << N, and can be decreased using small step sizes. This does not affect the properties of TunaMH.