Training custom dataset out of memory

[ttrungtin2910]

Hello everyone,
Follow this GitHub, I can train new model based on config: configs/cbnet/htc_cbv2_swin_large_patch4_window7_mstrain_400-1400_giou_4conv1f_adamw_1x_coco.py
and pre-trained model:
htc_cbv2_swin_large22k_patch4_window7_mstrain_400-1400_giou_4conv1f_adamw_1x_coco.pth

I use 2 RTX2080Ti 11GB but they're not enough memory, what should I do to can train my dataset

Or another suggests, please reply with me

Thanks very much

[taochen526]

same problem here