Closed robertnishihara closed 8 years ago
Hi Rahul, there is currently an issue in our example apps when using Caffe with GPUs. As seen in your comment toward the end of #100, the test accuracy goes way too high very quickly in ImageNetApp when using GPUs. The problem doesn't seem to happen with TensorFlow. Still looking into this...
The bug is fixed in #115, which essentially reverts this commit.
Hello Robert,
It will be great if you could let know if there is an issue with the GPU mode which we should be aware of, because of which it is being commented?
Thanks, Rahul