Pytorch Using nn.linear layer in 3d tensor

Question

I have an input tensor (num_data,channel,num_node)=(10,8,6) and used nn.Linear(6,1) to make output tensor (10,8,1). I wonder whether like general conv layer with groups=1, all channel output is concatenated to make new output or like general conv layer with groups=8, all channel output separably used make new output.