search

Haochen-Wang409 / U2PL

[CVPR'22] Semi-Supervised Semantic Segmentation Using Unreliable Pseudo-Labels
Apache License 2.0
426 stars 59 forks source link

Question about the GPU memory #124

Closed Hugo-cell111 closed 1 year ago

Hugo-cell111 commented 1 year ago

Hi! In the paper the batch size is set as 16 when training with Cityscapes, but when I just set the bs as 1, the memory has come to 16517 Mb. So I wonder how many GPUs are used when training with the batch size of 16? Looking forward to your reply. Thanks!

Haochen-Wang409 commented 1 year ago

The batch size for each GPU is 2 and we trained our model using 8 Tesla V100 (32G) GPUs.

Hugo-cell111 commented 1 year ago

Thanks!