NuTorch

Creation ops

Generated from the binaries by scripts/gen-ops-reference.ts — do not edit by hand. Every op also documents itself: torch <op> --help.

full

a tensor of the given shape filled with a value

torch full <shape> <value> [--dtype <Str>] [--requires_grad]

randn

standard-normal random tensor (float kinds only)

torch randn <shape> [--dtype <Str>] [--requires_grad]

zeros

a tensor of zeros

torch zeros <shape> [--dtype <Str>] [--requires_grad]

ones

a tensor of ones

torch ones <shape> [--dtype <Str>] [--requires_grad]

eye

identity matrix (n x n, or n x —m)

torch eye <n> [--m <Int>]

arange

range [—start, end) by —step (CLI reshape of PyTorch overloads)

torch arange <end> [--start <Scalar>] [--step <Scalar>]

linspace

steps evenly spaced points in [start, end]

torch linspace <start> <end> <steps>

rand

uniform [0,1) random tensor (seeded CPU generator)

torch rand <shape> [--requires_grad]

randint

random int64s in [—low, high) (seeded CPU generator)

torch randint <high> <shape> [--low <Int>]

zeros_like

zeros with the input’s shape and dtype

torch zeros_like <t1>

ones_like

ones with the input’s shape and dtype

torch ones_like <t1>

full_like

a value-filled tensor with the input’s shape and dtype

torch full_like <t1> <value>

rand_like

uniform random with the input’s shape (seeded CPU generator)

torch rand_like <t1>

randn_like

normal random with the input’s shape (seeded CPU generator)

torch randn_like <t1>