Facenet metadata

Hello.
I am working on Jetson AGX Xavier, working on deepstream 5.0. I working on generating the features embeddings. So my primary is from Nvidia transfer learning toolkit and my secondary is the Keras model.
The issue is after generating the metadata, the 128 features vector are approximately the same. It is too small in the range of 10^-7. I am not sure if I am probing the data correctly.

There is no update from you for a period, assuming this is not an issue any more.
Hence we are closing this topic. If need further support, please open a new one.
Thanks

Hey, customer
Can you confirm if the model can work well?