openspeech
openspeech copied to clipboard
Isn't Conv1D with kernel=1 the same as Linear layer but slower?
❓ Questions & Help
I was wandering whether it makes sense from an efficiency point of view to change the Conv1D layer with kernel 1 from the conformer encoder to standard linear layers. As I understand the operation is the same and a linear layer is ~10x faster.
Am I correct or am I missing something?
Thank you for the good points!
I looked it up, too. Please refer to https://stackoverflow.com/questions/55576314/conv1d-with-kernel-size-1-vs-linear-layer.