how about the result you got about coco?

[Wondersui]

No description provided.

[Angzz]

Hi, Wonder, currently we achieve 36.5 for ResNet-101 with 1.5x training strategy and single scale train, but recently I do not have time to complete this, thanks for your attention. Wonder <[email protected]> 于2019年5月13日周一 上午9:46写道：

…

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#1>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AIHPPCCPQS4SS7VYKDIHETLPVDB77ANCNFSM4HMLYHHQ> .