giagrad.Tensor.sparse

Tensor.sparse(sparsity, std=0.01) → Tensor

Fills the 2D Tensor data as a sparse matrix.

Non-zero elements will be drawn from the normal distribution \(\mathcal{N}(0, \text{sigma})\), as described in Deep learning via Hessian-free optimization - Martens, J. (2010).

Parameters:

• sparsity (float between [0, 1)) – The fraction of elements in each column to be set to zero.

• std (float) – The standard deviation of the normal distribution used to generate the non-zero values.

Examples

>>> Tensor.empty(3, 5).sparse(sparsity=0.4, std=0.2)