The following table lists the ai.onnx operators supported by each of the available backends.
See Compatibility for a list of the supported platforms.
| Operator | Cpu Backend | Wasm Backend | WebGl Backend |
|---|---|---|---|
| Abs | ✔️ | ✔️ | |
| Acos | ✔️ | ✔️ | |
| Add | ✔️ | ✔️ | |
| And | ✔️ | ✔️ | |
| Asin | ✔️ | ✔️ | |
| Atan | ✔️ | ✔️ | |
| AveragePool | ✔️ | ✔️ | ✔️ |
| BatchNormalization | ✔️ | ✔️ | ✔️ |
| Ceil | ✔️ | ✔️ | |
| Concat | ✔️ | ✔️ | |
| Constant | ✔️ | ✔️ | ✔️ |
| Conv | ✔️ | ✔️ | ✔️ |
| Cos | ✔️ | ✔️ | |
| Div | ✔️ | ✔️ | |
| Dropout | ✔️ | ✔️ | ✔️ |
| Equal | ✔️ | ||
| Exp | ✔️ | ✔️ | |
| Floor | ✔️ | ✔️ | |
| Gemm | ✔️ | ✔️ | |
| GlobalAveragePool | ✔️ | ✔️ | ✔️ |
| GlobalMaxPool | ✔️ | ✔️ | ✔️ |
| Greater | ✔️ | ||
| Identity | ✔️ | ✔️ | ✔️ |
| ImageScaler | ✔️ | ✔️ | |
| LeakyRelu | ✔️ | ✔️ | |
| Less | ✔️ | ||
| Log | ✔️ | ✔️ | |
| LRN | ✔️ | ||
| MatMul | ✔️ | ✔️ | |
| MaxPool | ✔️ | ✔️ | ✔️ |
| Mul | ✔️ | ✔️ | |
| Neg | ✔️ | ✔️ | |
| Not | ✔️ | ||
| Or | ✔️ | ✔️ | |
| Pad | ✔️ | ||
| Pow | ✔️ | ||
| PRelu | ✔️ | ✔️ | |
| ReduceLogSum | ✔️ | ||
| ReduceMax | ✔️ | ||
| ReduceMean | ✔️ | ||
| ReduceMin | |||
| ReduceProd | |||
| ReduceSum | |||
| ReduceSumSquare | |||
| Relu | ✔️ | ✔️ | |
| Reshape | ✔️ | ✔️ | |
| Sigmoid | ✔️ | ✔️ | |
| Sin | ✔️ | ✔️ | |
| Softmax | ✔️ | ✔️ | ✔️ |
| Split | ✔️ | ||
| Sqrt | ✔️ | ✔️ | |
| Sub | ✔️ | ✔️ | |
| Sum | ✔️ | ✔️ | ✔️ |
| Tan | ✔️ | ✔️ | |
| Tanh | ✔️ | ||
| Transpose | ✔️ | ✔️ | |
| Unsqueeze | ✔️ | ||
| Xor | ✔️ | ✔️ |