Open BW15061999 opened 3 years ago
Which graph size are you trying on?
I used command CUDA_VISIBLE_DEVICES=0 python train.py
in FINDER_CN_COST with BATCH_SIZE = 16
and others parameters remained untouched
What about graph size? How many graph nodes are you training with?
When I tried another deep learning program, the GPU memory has been occupied up to 90%. Is FINDER really not taking up a lot of GPU memory?