The paper clearly points out that achieving robustness under heavy-tailed noise requires a slightly stronger explicit sample size requirement than in the light-tailed case.