| --- |
| library_name: kernels |
| license: apache-2.0 |
| tags: |
| - kernel |
| - webgpu |
| - wgsl |
| --- |
| # ai.onnx.If |
|
|
| `ai.onnx` · internal tensor lowering (non-standard) · reviewed against ONNX opset 25 |
|
|
| ## Description |
|
|
| Support status: the standard ONNX `If` control-flow operator is not implemented because standalone kernel packages cannot carry or execute its `then_branch` and `else_branch` graph attributes. This internal lowering only selects elementwise between two pre-evaluated, equal-sized tensors from a scalar condition and must not be treated as ONNX `If`. |
|
|
| See the [standard ONNX `If` spec](https://onnx.ai/onnx/operators/onnx__If.html) for the contract this internal lowering does not implement. |
|
|
| ## Inputs |
|
|
| | Name | Bind key | Logical dtype | Rank | Shape | Description | Presence | |
| | --- | --- | --- | --- | --- | --- | --- | |
| | `cond` | `cond` | `B` | — | — | Scalar boolean condition that selects which value tensor to output. | required | |
| | `then_value` | `then_value` | `T` | — | — | Values to output when `cond` is true. | required | |
| | `else_value` | `else_value` | `T` | — | — | Values to output when `cond` is false. | required | |
|
|
| ## Outputs |
|
|
| | Name | Bind key | Logical dtype | Rank | Shape | Description | Presence | |
| | --- | --- | --- | --- | --- | --- | --- | |
| | `y` | `y` | `T` | — | — | Output tensor with the same number of elements as `then_value` and `else_value`. | required | |
|
|
| ## Type constraints |
|
|
| | Variable | Allowed dtypes | |
| | --- | --- | |
| | `T` | `float32` | |
| | `B` | `uint32`, `bool` | |
|
|
| ## Files |
|
|
| - [`metadata.json`](build/webgpu/metadata.json) — kernel metadata (id, digests, provenance) |
| - [`manifest.json`](build/webgpu/manifest.json) — the op contract (source of truth) |
| - [`test.json`](build/webgpu/test.json) — correctness cases |
| - [`bench.json`](build/webgpu/bench.json) — benchmark + tuning cases |
| - [`if-select.wgsl.jinja`](build/webgpu/if-select.wgsl.jinja) |
|
|
| ## Use with `@huggingface/kernels` |
|
|
| The loader automatically allocates outputs whose metadata it can derive from the manifest contract and this call. |
|
|
| The explicit `outputs` entries provide shape and logical dtype metadata for the results listed below: |
|
|
| - `y` |
|
|
| Each entry either requests an optional result or supplies metadata that cannot be inferred from the inputs. |
|
|
| The `version: 1` option selects the published kernel contract; it is independent of any operator opset, contrib `since_version`, or model version. |
|
|
| Replace each `*Data` placeholder with a typed array containing the corresponding input data. |
|
|
| ```js |
| import { getKernel } from "@huggingface/kernels"; |
| |
| const kernel = await getKernel("webgpu-kernels/ai.onnx.If", { version: 1 }); |
| // Explicit destinations request optional results or supply metadata that cannot be inferred. |
| const { y } = await kernel({ |
| cond: { data: condData, shape: [1] }, |
| then_value: { data: then_valueData, shape: [1] }, |
| else_value: { data: else_valueData, shape: [1] }, |
| }, { |
| outputs: { y: { shape: [1], dtype: "float32" } }, |
| }); |
| ``` |
|
|
