search

lxtGH / DecoupleSegNets

[ECCV-2020]: Improving Semantic Segmentation via Decoupled Body and Edge Supervision
370 stars 36 forks source link

GPU Usage #6

Closed chuzcjoe closed 4 years ago

chuzcjoe commented 4 years ago

Hi, i noticed that the whole network size is relatively small but when i started training the model, my GPU memory is almost fully loaded even i set the batch size to 4. I was wondering where the most computation and parameters come from. Thanks.

lxtGH commented 4 years ago

Hi what is your GPU memory? It has at least 12GB to run the training process.

chuzcjoe commented 4 years ago

Hi what is your GPU memory? It has at least 12GB to run the training process.

10GB

lxtGH commented 4 years ago

So you can not run this code with such limited GPU memory.