Problems on GPU and CPU usage

[jwfanDL]

Hi~ I meet some problems when training the model
I have set CUDA_VISIBEL_DEVICES=0,1,2,3, however, the code may just use gpu0.
At the same time, cpu usage rate is quite high.

So how can I use more GPU to train the model? I would appreciate if you can answer the question

[DongSky]

Hi JwFbupt
Currently, our LBBA only supports single GPU training, just like the description in readme "GPU with larger memory is better".
And now we are working on migrating the original official code into the multi-GPU detectron2 codebase, but it still needs some time due to some other things.
Best wishes.