Sphereface-model
Sphereface-model copied to clipboard
How to train on the larger dataset?
I have tried A-Softmax_loss on Ms-Celeb-1M datasets. But it is difficult to converge.Did you finished the experiments on Ms-Celeb-1M datasets.
I have not tried A-Softmax loss on Ms-Celeb-1M datasets. I would appreciated it if you shared your experiences about experiments on Ms-Celeb-1M datasets.