Could you please provide a pretrained model for the completion of 16384 resolution?

Hi Yida,
Thank you for your good work! Recently, I'm working on completing dense objects. I've noticed that softpoolnet has reached a performance of 5.94 much better than about 8~9 of former works on 16384 points completion. I think it is an outstanding and excellent work. So could you please provide a pretrained model for the completion of dense models like shown in table 1? 
