Moccasin: Efficient Tensor Rematerialization for Neural Networks

Open in new window