Closed ryanaleksander closed 5 years ago
@ryanaleksander The original SSD repo also suffers such problems. It seems that the training speed is sensitive to GPU and drivers. See https://github.com/weiliu89/caffe/issues/691 I guess the low gpu-util is caused by the data layer. You can open the debug mode and see the time consuming of each layer.
Ok, I fixed it. I just happened to use the same dataset for both training and testing so it didn't run
Hello. I'm using the docker image to train on my own dataset, but it's been stuck at the same place for hours, it's still stuck right now as I'm typing this. I don't know if it's even running properly or not since it's barely using any GPU memory.
Here are some screenshots