Hi, I am trying to use QuaternionicConv but I am not able to understand the input to the network. How the r, i , j , and k components of the data are given to the network.
As per my interpretation data should have four channels to represent the quaterionic numbers [[ r],[I],[j],[k]] -
So for Conv1d it becomes batchsize, input channel , sequence length where input_channel should be 4.
Pls, provide explanation how to represent data for your implementation QuaterionicConv and QuaterionicLinear.
Hi, I am trying to use QuaternionicConv but I am not able to understand the input to the network. How the r, i , j , and k components of the data are given to the network.
As per my interpretation data should have four channels to represent the quaterionic numbers [[ r],[I],[j],[k]] -
So for Conv1d it becomes batchsize, input channel , sequence length where input_channel should be 4.
Pls, provide explanation how to represent data for your implementation QuaterionicConv and QuaterionicLinear.