hey guys, since the encoder part of the UNet can be modified to a great classifier, I also used the encoder part of your UNode on a simple 2-category classification task, but the performances were very unstable while UNet-encoder reached 90%+ stably.
Could you provide any suggestions on this problem? Very appreciated!
hey guys, since the encoder part of the UNet can be modified to a great classifier, I also used the encoder part of your UNode on a simple 2-category classification task, but the performances were very unstable while UNet-encoder reached 90%+ stably.
Could you provide any suggestions on this problem? Very appreciated!