PyTorch Operators

Overview

This document summarizes the SynapseAI® Software PyTorch supported operators for Habana® Gaudi®. Note that the operators listed below support only selected variants and limited optional parameters for Gaudi.

PyTorch Operators Support Summary

PyTorch Operator

FP32

BF16

INT32

INT8

BOOL

torch.nn.functional

avg_pool2d

Yes

Yes

No

No

No

batch_norm

Yes

Yes

No

No

No

binary_cross_entropy

Yes

Yes

No

No

No

conv1d

Yes

Yes

No

No

No

conv2d

Yes

Yes

No

No

No

conv3d

Yes

Yes

No

No

No

dropout

Yes

Yes

No

No

No

embedding

Yes

Yes

No

No

No

embedding_bag

Yes

Yes

No

No

No

elu

Yes

Yes

No

No

No

elu_

Yes

Yes

No

No

No

gelu

Yes

Yes

No

No

No

glu

Yes

No

No

No

No

grid_sample

Yes

No

No

No

No

hardtanh

Yes

No

No

No

No

hardtanh_

Yes

No

No

No

No

hardsigmoid

Yes

Yes

No

No

No

huber_loss

Yes

Yes

No

No

No

kl_div

Yes

Yes

No

No

No

l1_loss

Yes

Yes

No

No

No

layer_norm

Yes

Yes

No

No

No

linear

Yes

Yes

No

No

No

log_softmax

Yes

Yes

No

No

No

max_pool2d

Yes

Yes

No

No

No

max_pool3d

Yes

Yes

No

No

No

mse_loss

Yes

Yes

No

No

No

nll_loss

Yes

Yes

No

No

No

pad

Yes

Yes

No

No

No

relu

Yes

Yes

No

No

No

sigmoid

Yes

Yes

No

No

No

tanh

Yes

Yes

No

No

No

upsample

Yes

Yes

No

No

No

upsample_nearest

Yes

Yes

No

No

No

binary_cross_entropy_with_logits

Yes

Yes

No

No

No

conv_transpose2d

Yes

Yes

No

No

No

instance_norm

Yes

Yes

No

No

No

silu

Yes

Yes

No

No

No

smooth_l1_loss

Yes

Yes

No

No

No

leaky_relu

Yes

Yes

No

No

No

mish

Yes

Yes

No

No

No

logsigmoid

Yes

No

No

No

No

threshold

Yes

Yes

No

No

No

rrelu

Yes

Yes

No

No

No

hardshrink

Yes

Yes

No

No

No

softshrink

Yes

Yes

No

No

No

relu

Yes

Yes

No

No

No

relu_

Yes

Yes

No

No

No

softmax

Yes

Yes

No

No

No

torch

abs

Yes

Yes

No

No

No

absolute

Yes

Yes

No

No

No

acos

Yes

No

No

No

No

acosh

Yes

No

No

No

No

add

Yes

Yes

No

No

No

addcdiv

Yes

Yes

No

No

No

addcmul

Yes

Yes

No

No

No

addmm

Yes

Yes

No

No

No

addmv

Yes

Yes

No

No

No

addr

Yes

Yes

No

No

No

all

No

No

No

No

Yes

amax

Yes

Yes

Yes

No

No

amin

Yes

Yes

Yes

No

No

aminmax

Yes

Yes

Yes

Yes

No

any

Yes

Yes

Yes

No

Yes

arange

Yes

Yes

Yes

No

No

arccos

Yes

No

No

No

No

arccosh

Yes

No

No

No

No

arcsin

Yes

No

No

No

No

arcsinh

Yes

No

No

No

No

arctan

Yes

No

No

No

No

arctanh

Yes

No

No

No

No

argmax

Yes

Yes

Yes

No

No

argmin

Yes

Yes

Yes

No

No

asin

Yes

No

No

No

No

asinh

Yes

No

No

No

No

atan

Yes

No

No

No

No

atanh

Yes

No

No

No

No

batched_nms

Yes

No

No

No

No

bernoulli

Yes

Yes

No

No

No

bmm

Yes

Yes

No

No

No

broadcast_tensors

Yes

Yes

No

No

No

cat

Yes

Yes

No

No

No

ceil

Yes

Yes

No

No

No

chunk

Yes

Yes

Yes

No

No

clip

Yes

Yes

Yes

No

No

conj

Yes

Yes

Yes

No

No

clone

Yes

Yes

Yes

Yes

Yes

copysign

Yes

Yes

No

No

No

cos

Yes

Yes

No

No

No

cosh

Yes

No

No

No

No

cross

Yes

Yes

Yes

No

No

cumprod

Yes

Yes

Yes

No

No

diag

Yes

Yes

No

No

No

div

Yes

Yes

Yes

Yes

No

divide

Yes

Yes

Yes

Yes

No

dot

Yes

Yes

No

No

No

empty

Yes

Yes

Yes

Yes

Yes

empty_like

Yes

Yes

Yes

Yes

Yes

empty_strided

Yes

Yes

Yes

Yes

Yes

eq

Yes

Yes

Yes

Yes

No

equal

Yes

Yes

Yes

Yes

Yes

erf

Yes

Yes

No

No

No

exp

Yes

Yes

No

No

No

exp2

Yes

Yes

No

No

No

expit

Yes

Yes

No

No

No

expm1

Yes

No

No

No

No

flatten

Yes

Yes

Yes

No

No

float_power

Yes

Yes

No

No

No

floor_divide

Yes

Yes

No

No

No

fmod

Yes

Yes

Yes

No

No

full

Yes

Yes

Yes

Yes

No

full_like

Yes

Yes

Yes

Yes

No

gather

Yes

Yes

Yes

Yes

No

gt

Yes

Yes

No

No

No

index_select

Yes

Yes

Yes

No

No

less

Yes

Yes

Yes

Yes

No

less_equal

Yes

Yes

Yes

Yes

No

linspace

Yes

Yes

No

No

No

log10

Yes

Yes

No

No

No

log1p

Yes

Yes

No

No

No

logcumsumexp

Yes

Yes

No

No

No

logit

Yes

Yes

No

No

No

logspace

Yes

Yes

No

No

No

logsumexp

Yes

Yes

No

No

No

lt

Yes

Yes

Yes

Yes

No

masked_select

Yes

Yes

Yes

No

No

matmul

Yes

Yes

No

No

No

mean

Yes

Yes

No

No

No

median

Yes

Yes

No

No

No

meshgrid

Yes

Yes

No

No

No

mul

Yes

Yes

No

No

No

mv

Yes

Yes

No

No

No

mm

Yes

Yes

No

No

No

nan_to_num

Yes

Yes

Yes

No

No

narrow

Yes

No

No

No

No

neg

Yes

Yes

No

No

No

norm

Yes

Yes

No

No

No

normal

Yes

Yes

No

No

No

not_equal

Yes

Yes

Yes

Yes

Yes

ones

Yes

Yes

Yes

Yes

Yes

ones_like

Yes

Yes

Yes

Yes

Yes

pow

Yes

Yes

No

No

No

random

Yes

Yes

Yes

No

No

reciprocal

Yes

Yes

No

No

No

resolve_conj

Yes

Yes

No

No

No

resolve_neg

Yes

Yes

No

No

No

result_type

Yes

Yes

Yes

No

No

roi_align

Yes

No

No

No

No

rsub

Yes

Yes

No

No

No

scatter

Yes

Yes

Yes

No

Yes

scatter_add

Yes

Yes

No

No

No

select

Yes

Yes

Yes

No

No

sigmoid

Yes

Yes

No

No

No

sin

Yes

Yes

No

No

No

sinc

Yes

Yes

No

No

No

sinh

Yes

No

No

No

No

sort

Yes

Yes

No

No

No

sqrt

Yes

Yes

No

No

No

square

Yes

Yes

Yes

No

No

squeeze

Yes

Yes

Yes

No

No

stack

Yes

Yes

Yes

No

No

sub

Yes

Yes

No

No

No

sum

Yes

Yes

Yes

No

Yes

tan

Yes

No

No

No

No

tanh

Yes

Yes

No

No

No

topk

Yes

Yes

No

No

No

transpose

Yes

Yes

Yes

Yes

Yes

tril

Yes

Yes

No

No

No

triu

Yes

Yes

No

No

No

trunc

Yes

Yes

No

No

No

unbind

Yes

Yes

Yes

No

No

uniform

Yes

Yes

No

No

No

unsqueeze

Yes

Yes

Yes

No

No

var_mean

Yes

Yes

No

No

No

zeros

Yes

Yes

Yes

Yes

Yes

zeros_like

Yes

Yes

Yes

Yes

Yes

max

Yes

Yes

Yes

No

No

maximum

Yes

Yes

Yes

No

No

minimum

Yes

Yes

Yes

No

No

floor

Yes

Yes

No

No

No

rsqrt

Yes

Yes

No

No

No

log

Yes

Yes

No

No

No

log2

Yes

Yes

No

No

No

round

Yes

Yes

No

No

No

slice

Yes

Yes

No

No

No

where

Yes

Yes

Yes

No

No

nansum

Yes

No

No

No

No

nms

Yes

No

No

No

No

ne

Yes

Yes

Yes

No

No

ge

Yes

Yes

Yes

No

No

all

No

No

No

No

Yes

sgn

Yes

Yes

No

No

No

sign

Yes

Yes

No

No

No

bitwise_and

No

No

Yes

Yes

Yes

bitwise_left_shift

Yes

Yes

Yes

Yes

Yes

bitwise_not

No

No

No

No

Yes

bitwise_or

No

No

Yes

Yes

Yes

bitwise_right_shift

Yes

Yes

Yes

Yes

Yes

bitwise_xor

No

No

Yes

Yes

Yes

prod

Yes

Yes

No

No

No

randperm

No

No

Yes

No

No

isfinite

Yes

Yes

No

No

No

nonzero

Yes

Yes

Yes

No

Yes

argmax

Yes

Yes

No

No

No

unique

Yes

Yes

No

No

No

clamp

Yes

Yes

Yes

No

No

reshape

Yes

Yes

Yes

Yes

Yes

flip

Yes

Yes

Yes

Yes

Yes

le

Yes

Yes

Yes

Yes

No

min

Yes

Yes

Yes

No

No

remainder

Yes

No

Yes

Yes

No

randperm

No

No

Yes

No

No

cumsum

Yes

Yes

Yes

No

No

equal

Yes

Yes

Yes

Yes

No

erfc

Yes

Yes

No

No

No

eye

Yes

Yes

No

No

No

xlogy

Yes

Yes

No

No

No

frac

Yes

Yes

No

No

No

greater

Yes

Yes

Yes

Yes

No

greater_equal

Yes

Yes

Yes

Yes

No

heaviside

Yes

Yes

Yes

No

No

hypot

Yes

Yes

No

No

No

is_complex

Yes

Yes

Yes

No

No

is_floating_point

Yes

Yes

Yes

No

No

is_nonzero

Yes

Yes

Yes

No

No

isnan

Yes

Yes

Yes

No

No

lerp

Yes

Yes

Yes

No

No

logaddexp

Yes

Yes

No

No

No

logaddexp2

Yes

Yes

No

No

No

signbit

Yes

Yes

Yes

Yes

No

trace

Yes

Yes

Yes

No

No

vdot

Yes

Yes

No

No

No

take

Yes

Yes

Yes

No

No

logical_and

Yes

Yes

No

Yes

No

logical_or

Yes

Yes

No

Yes

No

logical_xor

Yes

Yes

No

Yes

No

split_with_sizes

Yes

Yes

Yes

No

No

t

Yes

Yes

Yes

Yes

Yes

var

Yes

Yes

No

No

No

torch.nn

AvgPool2d

Yes

Yes

No

No

No

BatchNorm1d

Yes

Yes

No

No

No

BatchNorm2d

Yes

Yes

No

No

No

BCELoss

Yes

Yes

No

No

No

ChannelShuffle

Yes

Yes

Yes

No

No

CrossEntropyLoss

Yes

Yes

No

No

No

Hardtanh

Yes

No

No

No

No

ReLU

Yes

Yes

No

No

No

ReplicationPad1d

Yes

Yes

Yes

No

No

ReplicationPad2d

Yes

Yes

Yes

No

No

ReplicationPad3d

Yes

Yes

Yes

No

No

Upsample

Yes

Yes

No

No

No

UpsamplingNearest2d

Yes

Yes

No

No

No

BCEWithLogitsLoss

Yes

Yes

No

No

No

SiLU

Yes

Yes

No

No

No

SmoothL1Loss

Yes

Yes

No

No

No

LeakyReLU

Yes

Yes

No

No

No

ReflectionPad1d

Yes

No

No

No

No

ReflectionPad2d

Yes

No

No

No

No

ReflectionPad3d

Yes

No

No

No

No

Mish

Yes

Yes

No

No

No

LogSigmoid

Yes

No

No

No

No

Threshold

Yes

Yes

No

No

No

RReLU

Yes

Yes

No

No

No

Hardshrink

Yes

Yes

No

No

No

Softshrink

Yes

Yes

No

No

No

ConstantPad1d

Yes

Yes

No

No

No

Conv1d

Yes

Yes

No

No

No

Conv2d

Yes

Yes

No

No

No

Conv3d

Yes

Yes

No

No

No

Softmax

Yes

Yes

No

No

No

MSELoss

Yes

Yes

No

No

No

NLLLoss

Yes

Yes

No

No

No

Hardsigmoid

Yes

Yes

No

No

No

HuberLoss

Yes

Yes

No

No

No

InstanceNorm2d

Yes

Yes

No

No

No

Dropout

Yes

Yes

No

No

No

Embedding

Yes

Yes

No

No

No

EmbeddingBag

Yes

Yes

No

No

No

ConvTranspose2d

Yes

Yes

No

No

No

ConvTranspose3d

Yes

Yes

No

No

No

ELU

Yes

Yes

No

No

No

GELU

Yes

Yes

No

No

No

GLU

Yes

No

No

No

No

KLDivLoss

Yes

Yes

No

No

No

LayerNorm

Yes

Yes

No

No

No

Linear

Yes

Yes

No

No

No

LogSoftmax

Yes

Yes

No

No

No

MaxPool2d

Yes

Yes

No

No

No

MaxPool3d

Yes

Yes

No

No

No

torch.nn.utils

weight_norm

Yes

Yes

No

No

No

torch.Tensor

__and__

No

No

Yes

Yes

Yes

__iand__

No

No

Yes

Yes

Yes

__ilshift__

Yes

Yes

Yes

Yes

Yes

__ior__

No

No

Yes

Yes

Yes

__irshift__

Yes

Yes

Yes

Yes

Yes

__ixor__

No

No

Yes

Yes

Yes

__lshift__

Yes

Yes

Yes

Yes

Yes

__or__

No

No

Yes

Yes

Yes

__rshift__

Yes

Yes

Yes

Yes

Yes

__xor__

No

No

Yes

Yes

Yes

abs

Yes

Yes

No

No

No

acos

Yes

No

No

No

No

acos_

Yes

No

No

No

No

acosh

Yes

No

No

No

No

acosh_

Yes

No

No

No

No

add

Yes

Yes

No

No

No

add_

Yes

Yes

No

No

No

addcdiv

Yes

Yes

No

No

No

addcdiv_

Yes

Yes

No

No

No

addcmul

Yes

Yes

No

No

No

addcmul_

Yes

Yes

No

No

No

addmm

Yes

Yes

No

No

No

addmm_

Yes

Yes

No

No

No

addmv

Yes

Yes

No

No

No

addmv_

Yes

Yes

No

No

No

addr

Yes

Yes

No

No

No

addr_

Yes

Yes

No

No

No

all

No

No

No

No

Yes

amax

Yes

Yes

No

No

No

amin

Yes

Yes

No

No

No

aminmax

Yes

Yes

Yes

Yes

No

any

No

No

No

No

Yes

arccos

Yes

No

No

No

No

arccos_

Yes

No

No

No

No

arccosh

Yes

No

No

No

No

arccosh_

Yes

No

No

No

No

arcsin

Yes

No

No

No

No

arcsin_

Yes

No

No

No

No

arcsinh

Yes

No

No

No

No

arcsinh_

Yes

No

No

No

No

arctan

Yes

No

No

No

No

arctan_

Yes

No

No

No

No

arctanh

Yes

No

No

No

No

arctanh_

Yes

No

No

No

No

asin

Yes

No

No

No

No

asin_

Yes

No

No

No

No

asinh

Yes

No

No

No

No

asinh_

Yes

No

No

No

No

atanh

Yes

No

No

No

No

atanh_

Yes

No

No

No

No

bernoulli

Yes

Yes

No

No

No

bmm

Yes

Yes

No

No

No

ceil

Yes

Yes

No

No

No

clamp

Yes

Yes

Yes

No

No

clamp_

Yes

Yes

Yes

No

No

clamp_max

Yes

Yes

Yes

No

No

clamp_min

Yes

Yes

Yes

No

No

clip

Yes

Yes

Yes

No

No

clip_

Yes

Yes

Yes

No

No

clone

Yes

Yes

Yes

Yes

Yes

conj

Yes

Yes

Yes

No

No

copy_

Yes

Yes

Yes

Yes

Yes

copysign

Yes

Yes

No

No

No

copysign_

Yes

Yes

No

No

No

cos

Yes

Yes

No

No

No

cos_

Yes

Yes

No

No

No

cosh_

Yes

Yes

No

No

No

cross

Yes

Yes

Yes

No

No

cumprod

Yes

Yes

Yes

No

No

diag

Yes

Yes

No

No

No

div

Yes

Yes

Yes

Yes

No

div_

Yes

Yes

Yes

Yes

No

dot

Yes

Yes

No

No

No

fill_

Yes

Yes

Yes

No

No

eq

Yes

Yes

Yes

Yes

No

eq_

Yes

Yes

Yes

Yes

No

erf

Yes

Yes

No

No

No

erf_

Yes

Yes

No

No

No

exp

Yes

Yes

No

No

No

exp_

Yes

Yes

No

No

No

expand

Yes

Yes

Yes

No

No

expand_as

Yes

Yes

Yes

Yes

Yes

expm1

Yes

No

No

No

No

expm1_

Yes

No

No

No

No

exponential_

Yes

Yes

No

No

No

flatten

Yes

Yes

Yes

No

No

floor_divide

Yes

Yes

Yes

No

No

fmod

Yes

Yes

Yes

No

No

fmod_

Yes

Yes

Yes

No

No

gather

Yes

Yes

No

No

No

gt

Yes

Yes

Yes

Yes

No

index_add_

Yes

Yes

yes

No

No

index_put

Yes

Yes

Yes

No

No

index_put_

Yes

Yes

Yes

No

No

index_select

Yes

Yes

Yes

No

No

isnan

Yes

Yes

Yes

No

No

is_complex

Yes

Yes

Yes

No

No

is_floating_point

Yes

Yes

Yes

No

No

item

Yes

No

No

No

No

less

Yes

Yes

Yes

Yes

No

less_

Yes

Yes

Yes

Yes

No

less_equal

Yes

Yes

Yes

Yes

No

less_equal_

Yes

Yes

Yes

Yes

No

log_softmax

Yes

Yes

No

No

No

log10

Yes

Yes

No

No

No

log1p

Yes

Yes

No

No

No

log1p_

Yes

Yes

No

No

No

logcumsumexp

Yes

Yes

No

No

No

logit

Yes

Yes

No

No

No

logsumexp

Yes

Yes

No

No

No

lt

Yes

Yes

Yes

Yes

No

masked_fill

Yes

Yes

Yes

No

No

masked_fill_

Yes

Yes

Yes

No

No

masked_select

Yes

Yes

Yes

No

No

matmul

Yes

Yes

No

No

No

mean

Yes

Yes

No

No

No

median

Yes

Yes

No

No

No

mm

Yes

Yes

No

No

No

mul

Yes

Yes

No

No

No

mul_

Yes

Yes

No

No

No

mv

Yes

Yes

No

No

No

nan_to_num

Yes

Yes

Yes

No

No

nansum

Yes

No

No

No

No

narrow

Yes

No

No

No

No

neg

Yes

Yes

No

No

No

new_empty

Yes

Yes

Yes

Yes

No

new_empty_strided

Yes

Yes

Yes

Yes

No

new_full

Yes

Yes

Yes

Yes

No

new_ones

Yes

Yes

Yes

Yes

No

new_zeros

Yes

Yes

Yes

Yes

No

norm

Yes

Yes

No

No

No

normal_

Yes

Yes

No

No

No

permute

Yes

Yes

Yes

No

No

pow

Yes

Yes

No

No

No

pow_

Yes

Yes

No

No

No

reciprocal

Yes

Yes

No

No

No

reciprocal_

Yes

Yes

No

No

No

scatter_

Yes

Yes

No

No

No

scatter_add

Yes

Yes

No

No

No

scatter_add_

Yes

Yes

No

No

No

select

Yes

Yes

Yes

No

No

sigmoid

Yes

Yes

No

No

No

sinh

Yes

No

No

No

No

sinh_

Yes

No

No

No

No

sigmoid_

Yes

Yes

No

No

No

remainder_

No

No

Yes

Yes

No

sort

Yes

Yes

No

No

No

split_with_sizes

Yes

Yes

Yes

No

No

sqrt

Yes

Yes

No

No

No

square

Yes

Yes

No

No

No

square_

Yes

Yes

No

No

No

squeeze

Yes

Yes

Yes

No

No

sub

Yes

Yes

No

No

No

sub_

Yes

Yes

No

No

No

sum

Yes

Yes

No

No

No

T

Yes

Yes

Yes

Yes

Yes

t

Yes

Yes

Yes

Yes

Yes

tan

Yes

No

No

No

No

tan_

Yes

No

No

No

No

tanh

Yes

Yes

No

No

No

tanh_

Yes

Yes

No

No

No

to

Yes

Yes

Yes

Yes

Yes

topk

Yes

Yes

No

No

No

transpose

Yes

Yes

Yes

Yes

Yes

tril

Yes

Yes

No

No

No

tril_

Yes

Yes

No

No

No

triu

Yes

Yes

No

No

No

triu_

Yes

Yes

No

No

No

trunc

Yes

Yes

No

No

No

trunc_

Yes

Yes

No

No

No

unbind

Yes

Yes

Yes

No

No

unsqueeze

Yes

Yes

Yes

No

No

uniform_

Yes

Yes

No

No

No

view

Yes

No

No

No

No

max

Yes

Yes

No

No

No

maximum

Yes

Yes

Yes

No

No

minimum

Yes

Yes

Yes

No

No

floor

Yes

Yes

No

No

No

floor_

Yes

Yes

No

No

No

rsqrt

Yes

Yes

No

No

No

rsqrt_

Yes

Yes

No

No

No

log

Yes

Yes

No

No

No

log2

Yes

Yes

No

No

No

log2_

Yes

Yes

No

No

No

log_

Yes

Yes

No

No

No

round

Yes

Yes

No

No

No

round_

Yes

Yes

No

No

No

where

Yes

Yes

Yes

No

No

ne

Yes

Yes

Yes

No

No

ne_

Yes

Yes

Yes

No

No

ge

Yes

Yes

Yes

No

No

ge_

Yes

Yes

Yes

No

No

all

No

No

No

No

Yes

sgn

Yes

Yes

No

No

No

sgn_

Yes

Yes

No

No

No

sign

Yes

Yes

No

No

No

sign_

Yes

Yes

No

No

No

bitwise_and

No

No

No

No

Yes

bitwise_left_shift

Yes

Yes

Yes

Yes

No

bitwise_not

No

No

Yes

Yes

Yes

bitwise_or

No

No

No

No

Yes

bitwise_right_shift

Yes

Yes

Yes

Yes

No

bitwise_xor

No

No

No

No

Yes

prod

Yes

Yes

No

No

No

isfinite

Yes

Yes

No

No

No

nonzero

Yes

Yes

Yes

No

Yes

argmax

Yes

Yes

No

No

No

unique

Yes

Yes

No

No

No

var

Yes

Yes

No

No

No

reshape

Yes

Yes

Yes

Yes

Yes

flip

Yes

Yes

Yes

Yes

Yes

le

Yes

Yes

Yes

Yes

No

min

Yes

Yes

Yes

No

No

pin_memory

Yes

Yes

Yes

Yes

Yes

remainder

Yes

No

Yes

Yes

No

remainder_

No

No

Yes

Yes

No

cos

Yes

Yes

No

No

No

cos_

Yes

Yes

No

No

No

sin

Yes

Yes

No

No

No

sin_

Yes

Yes

No

No

No

sinc

Yes

Yes

No

No

No

cumsum

Yes

Yes

Yes

No

No

repeat

Yes

Yes

Yes

Yes

Yes

le_

Yes

Yes

Yes

Yes

No

equal

Yes

Yes

Yes

Yes

No

erfc

Yes

Yes

No

No

No

frac

Yes

Yes

No

No

No

greater

Yes

Yes

Yes

Yes

No

greater_equal

Yes

Yes

Yes

Yes

No

heaviside

Yes

Yes

Yes

No

No

hypot

Yes

Yes

No

No

No

lerp

Yes

Yes

Yes

No

No

logaddexp

Yes

Yes

No

No

No

logaddexp2

Yes

Yes

No

No

No

signbit

Yes

Yes

Yes

Yes

No

trace

Yes

Yes

Yes

No

No

xlogy

Yes

Yes

No

No

No

xlogy_

Yes

Yes

No

No

No

vdot

Yes

Yes

No

No

No

take

Yes

Yes

Yes

No

No

logical_and

Yes

Yes

No

Yes

No

logical_or

Yes

Yes

No

Yes

No

logical_xor

Yes

Yes

No

Yes

No

hardshrink

Yes

Yes

No

No

No

geometric_

Yes

Yes

No

No

No

torch.special

log1p

Yes

Yes

No

No

No

logit

Yes

Yes

No

No

No

logsumexp

Yes

Yes

No

No

No

xlogy

Yes

Yes

No

No

No

xlog1py

Yes

Yes

No

No

No

entr

Yes

No

No

No

No

expit

Yes

Yes

No

No

No

expm1

Yes

No

No

No

No