Reproduced the results on D-Wave 2000Q and Advantage systems for 1000 samples and 1 shot each, using 20 us annealing time. Despite some (expected) statistical variations, roughly the same result was obtained. However, it is not clear whether the reverse annealing is reinitialized at each sampling. Without reinitialization, a much lower success rate of finding the g.s. is realized, and in pegasus topology, the solution is excited far away from the g.s., with zero probability of finding the g.s.