HI
Now I train my model with batch-size, BATCH_SIZE=2 for example.
And in each batch, there is an operation: A Matirx Mul A Vector:
So the batch matrix’s size can be (2, 3,4)
and then the batch vector size can be (2, 1, 3)
in each batch , the matrix mul will be executed: (1,3) X (3,4), which get a vector( 1,4)
So with batch size, How can I get the final tensor with size : (2, 1, 4) . And torch.mul does not work