search

bearpaw / pose-attention

Code for "Multi-Context Attention for Human Pose Estimation " (CVPR 2017)
BSD 3-Clause "New" or "Revised" License
170 stars 49 forks source link

Out of memory problem #10

Closed wnzhyee closed 6 years ago

wnzhyee commented 7 years ago

I see that in train.sh, it use 4x12G GPUs for training, but i only have one 12G GPU(1080 Ti), so how can i reduce the GPU cost?

i try change '-nGPU 4' to '-nGPU 1', and add '-trainBatch 4' in train.sh, but it seems doesn't work because the model 'converting the model to CUDA' again and again, just like:

_2017-11-07_21-39-01

So how can i do?Thanks.

bearpaw commented 6 years ago

This might not be an 'out of memory' issue. We've seen this occasionally also in our experiment, and we don't know the reason.

You can simply kill the program and rerun it.

larklili commented 6 years ago

@wnzhyee i have only one 12GB gpu too.what's your solution?could you please tell me the method?:)thanks