SUPPORTED-ONNX-OPS.md

Supported ONNX Operators

This table lists the support status for ONNX operators in Burn. Note that some entries marked with dimensional suffixes (such as Conv1d, Conv2d, etc.) or other specialized names like Linear are not standard ONNX operators. These represent Burn's implementation of dimension-specific versions of the corresponding ONNX operators to make the mapping clearer between ONNX and Burn functionality.

ONNX OP	Import Support	Burn Support
Abs	✅	✅
Acos	❌	❌
Acosh	❌	❌
Add	✅	✅
And	❌	✅
ArgMax	✅	✅
ArgMin	❌	✅
Asin	❌	❌
Asinh	❌	❌
Atan	❌	❌
Atanh	❌	❌
Attention	❌	✅
AveragePool1d	✅	✅
AveragePool2d	✅	✅
BatchNormalization	✅	✅
Bernoulli	❌	❌
BitShift	❌	✅
BitwiseAnd	❌	✅
BitwiseNot	❌	✅
BitwiseOr	❌	✅
BitwiseXor	❌	✅
BlackmanWindow	❌	❌
Cast	✅	✅
CastLike	❌	❌
Ceil	❌	✅
Celu	❌	❌
CenterCropPad	❌	❌
Clip	✅	✅
Col2Im	❌	❌
Compress	❌	❌
Concat	✅	✅
ConcatFromSequence	❌	❌
Constant	✅	✅
ConstantOfShape	✅	✅
Conv1d	✅	✅
Conv2d	✅	✅
Conv3d	✅	✅
ConvInteger	❌	❌
ConvTranspose1d	✅	✅
ConvTranspose2d	✅	✅
ConvTranspose3d	✅	✅
Cos	✅	✅
Cosh	✅	✅
CumSum	❌	❌
DeformConv	❌	❌
DepthToSpace	❌	❌
DequantizeLinear	❌	❌
Det	❌	❌
DFT	❌	❌
Div	✅	✅
Dropout	✅	✅
DynamicQuantizeLinear	❌	❌
Einsum	❌	❌
Elu	❌	❌
Equal	✅	✅
Erf	✅	✅
Exp	✅	✅
Expand	✅	✅
EyeLike	❌	✅
Flatten	✅	✅
Floor	✅	✅
Gather	✅	✅
GatherElements	✅	✅
GatherND	❌	❌
Gelu	✅	✅
Gemm	✅	✅
GlobalAveragePool	✅	✅
GlobalLpPool	❌	❌
GlobalMaxPool	❌	❌
Greater	✅	✅
GreaterOrEqual	✅	✅
GridSample	❌	❌
GroupNormalization	❌	✅
GRU	❌	✅
HammingWindow	❌	❌
HannWindow	❌	❌
Hardmax	❌	❌
HardSigmoid	✅	✅
HardSwish	❌	❌
Identity	✅	✅
If	❌	✅
Im	❌	❌
InstanceNormalization	❌	✅
IsInf	❌	❌
IsNaN	❌	❌
LayerNormalization	✅	✅
LeakyRelu	✅	✅
Less	✅	✅
LessOrEqual	✅	✅
Linear	✅	✅
Log	✅	✅
LogSoftmax	✅	✅
Loop	❌	❌
LpNormalization	❌	❌
LpPool	❌	❌
LRN	❌	❌
LSTM	❌	✅
MatMul	✅	✅
MatMulInteger	❌	✅
Max	✅	✅
MaxPool1d	✅	✅
MaxPool2d	✅	✅
MaxRoiPool	❌	❌
MaxUnpool	❌	❌
Mean	✅	✅
MeanVarianceNormalization	❌	❌
MelWeightMatrix	❌	❌
Min	✅	✅
Mish	❌	❌
Mod	❌	❌
Mul	✅	✅
Multinomial	❌	❌
Neg	✅	✅
NegativeLogLikelihoodLoss	❌	❌
NonMaxSuppression	❌	❌
NonZero	❌	✅
Not	✅	✅
OneHot	✅	✅
Optional	❌	❌
OptionalGetElement	❌	❌
OptionalHasElement	❌	❌
Or	❌	✅
Pad	✅	✅
Pow	✅	✅
PRelu	✅	✅
[QLinearConv][123]	❌	❌
[QLinearMatMul][124]	❌	❌
QuantizeLinear	❌	❌
RandomNormal	✅	✅
RandomNormalLike	✅	✅
RandomUniform	✅	✅
RandomUniformLike	✅	✅
Range	✅	✅
Reciprocal	✅	✅
ReduceL	❌	❌
ReduceLogSum	❌	❌
ReduceLogSumExp	❌	❌
ReduceMax	✅	✅
ReduceMean	✅	✅
ReduceMin	✅	✅
ReduceProd	✅	✅
ReduceSum	✅	✅
ReduceSumSquare	❌	❌
Relu	✅	✅
Reshape	✅	✅
Resize	✅	✅
ReverseSequence	❌	❌
RNN	❌	✅
RoiAlign	❌	❌
Round	❌	✅
Scan	❌	❌
Scatter	❌	✅
ScatterElements	❌	❌
ScatterND	❌	❌
Selu	❌	❌
SequenceAt	❌	❌
SequenceConstruct	❌	❌
SequenceEmpty	❌	❌
SequenceErase	❌	❌
SequenceInsert	❌	❌
SequenceLength	❌	❌
SequenceMap	❌	❌
Shape	✅	✅
Shrink	❌	❌
Sigmoid	✅	✅
Sign	✅	✅
Sin	✅	✅
Sinh	✅	✅
Size	❌	❌
Slice	✅	✅
Softmax	✅	✅
SoftmaxCrossEntropyLoss	❌	❌
Softplus	❌	❌
Softsign	❌	❌
SpaceToDepth	❌	❌
Split	✅	✅
SplitToSequence	❌	❌
Sqrt	✅	✅
Squeeze	✅	✅
STFT	❌	❌
StringNormalizer	❌	❌
Sub	✅	✅
Sum	✅	✅
Tan	✅	✅
Tanh	✅	✅
TfIdfVectorizer	❌	❌
ThresholdedRelu	❌	❌
Tile	✅	✅
TopK	✅	✅
Transpose	✅	✅
Trilu	✅	✅
Unique	❌	❌
Upsample	❌	❌
Where	✅	✅
Xor	❌	❌
Unsqueeze	✅	✅