hi,
I‘m learning to pvanet, using the caffe of pvanet version to other task. In the testing process, I'm using multi-scale tests, reporting 'out of memory'. Through to compare with the origin caffe, caffe of pvanet version take up more GPU memory. according to scale increase, the task cost more and more GPU memory until report errors. So, Whether or not there is optimization on the caffe of pvanet version about gpu memory in running time? can you give me some advices? thank you!
hi, I‘m learning to pvanet, using the caffe of pvanet version to other task. In the testing process, I'm using multi-scale tests, reporting 'out of memory'. Through to compare with the origin caffe, caffe of pvanet version take up more GPU memory. according to scale increase, the task cost more and more GPU memory until report errors. So, Whether or not there is optimization on the caffe of pvanet version about gpu memory in running time? can you give me some advices? thank you!