Speeding-up about training

[WWWJL]

Hello, has TensorNet accelerated during the training ? You only mentioned that The TT-FC layer is also faster than its uncompressed analog, so the TT-conv layer is not speeding up?Hope you can share the results of the training speed, thank you