soundnet
soundnet copied to clipboard
All features negative
hi, i used trained model to extracti feature from mp3s, but all 1000 dimentional features are negative. is this normal?
I also get the same result, all features are negative