Results of GPS for COCO-SP

[siqim]

Hi,

Just a quick question: How did you get the results of GPS for the COCO-SP dataset? It seems that the paper of GPS reports 0.3412 F1, but you report it with 0.3774 F1.

Thanks,
Siqi