OOM when training SeqFormer_swin_L on YouTube-VIS 2019 and COCO
JiaDingCN opened this issue · comments
jiading commented
Hi, thank you for your interesting work! I was trying to run your code but I meet OOM when training SeqFormer_swin_L on YouTube-VIS 2019 and COCO by your given script and command. I use 2 nodes and each node contains 8 V100 cards. Did I do something wrong?
Zhengkai Jiang commented
@wjf5203 meet the same problem
Junfeng Wu commented
The scales and max_size here should be set to [288, 320, 352, 392, 416] and 640 for swin_L backbone.
Line 208 in edbfba4