The matrix multiplication is only deterministic for sparse-dense products under these settings: > torch.bmm() when called on sparse-dense CUDA tensors And it'

The matrix multiplication is only deterministic for sparse-dense products under these settings: > torch.bmm() when called on sparse-dense CUDA tensors And it's not listed under the operations that raise an exception otherwise, so I'm not sure the docs promise that dense-dense matrix-matrix products are deterministic.