Hi, what is the best performance (hits@10, mean rank) for each model in your experiments? Could you please share the results and settings?