or prager

you can transform all train tfrecords labels to your labels and then re-train willow

@punitagrawal32 I can tell that I made a test ,In witch I used audioset for audio feature extraction and google frame feature extractor. I tested the accuracy of the willow...

I used audioset embedding op (tensor) with the default properties On Tue, Nov 6, 2018, 2:20 PM estathop did you choose little-endian or big-endian for the audioset feature > extraction...