xlnstorch.csrc.matmul_backward#
- xlnstorch.csrc.matmul_backward(grad, A, B, base)#
Perform the backward pass for LNS matrix multiplication.
- Parameters:
grad (torch.Tensor) – The gradient tensor in LNS format (torch.int64).
A (torch.Tensor) – The first tensor in LNS format (torch.int64).
B (torch.Tensor) – The second tensor in LNS format (torch.int64).
base (torch.Tensor) – The base for the LNS representation.
- Returns:
The output gradient tensor in LNS format (torch.int64).
- Return type: