Supported onnx operators

February 19, 2021 ยท View on GitHub

These are the onnx operators that are supported by tensor-js at the moment. If you'd like to see more, open an issue or a Pull Request if you have it implemented already. If not specified otherwise, all operators are supported on all backends and on onnx opsets 7 - 12.

This file is automatically generated. To update it run

 $ npm run onnxTable
OperatorOpsetsBackends
AbsAllAll
AcosAllAll
Acosh9,10,11,12CPU,WASM
AddAllAll
AsinAllAll
Asinh9,10,11,12CPU,WASM
AtanAllAll
Atanh9,10,11,12CPU,WASM
BatchNormalization9,10All
CeilAllAll
Celu12All
Clip7,8,9,10All
ConcatAllAll
ConstantAllAll
ConstantOfShape9,10,11,12All
ConvAllAll
CosAllAll
Cosh9,10,11,12All
DivAllAll
EluAllAll
ExpAllAll
Expand9,10,11,12All
FlattenAllAll
FloorAllAll
GatherAllAll
GemmAllAll
GlobalAveragePoolAllAll
HardSigmoidAllAll
IdentityAllAll
InstanceNormalization9,10,11,12All
LeakyReluAllAll
LogAllAll
MatMulAllAll
MeanAllAll
MulAllAll
NegAllAll
PReluAllAll
Pad7,8,9,10All
PowAllAll
Range11,12All
ReciprocalAllAll
ReduceL1AllAll
ReduceL2AllAll
ReduceLogSumAllAll
ReduceLogSumExpAllAll
ReduceMaxAllAll
ReduceMeanAllAll
ReduceMinAllAll
ReduceProdAllAll
ReduceSumAllAll
ReduceSumSquareAllAll
ReluAllAll
ReshapeAllAll
Round11,12All
SeluAllAll
ShapeAllAll
SigmoidAllAll
Sign9,10,11,12All
SinAllAll
Sinh9,10,11,12All
SizeAllAll
SliceAllAll
SoftmaxAllAll
SoftplusAllAll
SoftsignAllAll
SqrtAllAll
SqueezeAllAll
SubAllAll
SumAllAll
TanAllAll
Tanh9,10,11,12All
TileAllAll
TransposeAllAll
Unsqueeze7,8,9,10All
UpsampleAllAll