| --- |
| library_name: kernels |
| license: apache-2.0 |
| tags: |
| - kernel |
| - webgpu |
| - wgsl |
| --- |
| # com.microsoft.QuickGelu |
|
|
| `com.microsoft` · ONNX Runtime contrib operator · contrib since_version 1 |
| |
| ## Description |
| |
| Computes `x * sigmoid(alpha * x)` elementwise, a fast approximation of GELU activation. The output has the same shape as the input. This WebGPU package implements float16 and float32; the schema-allowed double and bfloat16 types are not supported. |
| |
| See the [ONNX Runtime `QuickGelu` contrib-operator spec](https://github.com/microsoft/onnxruntime/blob/main/docs/ContribOperators.md#com.microsoft.QuickGelu) for the reference semantics. |
| |
| ## Inputs |
| |
| | Name | Logical dtype | Rank | Shape | Description | Presence | |
| | --- | --- | --- | --- | --- | --- | |
| | `X` | `T` | — | — | Input tensor of any shape. | required | |
| |
| ## Outputs |
| |
| | Name | Logical dtype | Rank | Shape | Description | Presence | |
| | --- | --- | --- | --- | --- | --- | |
| | `Y` | `T` | same as `X` | same as `X` | Output tensor; same shape as the input. | required | |
| |
| ## Attributes |
| |
| Default values (overridable per request): |
| |
| | Attribute | Default | Description | |
| | --- | --- | --- | |
| | `alpha` | `1.702` | Scalar multiplier applied to `x` inside the sigmoid; defaults to 1.702, which approximates GELU. | |
| |
| ## Type constraints |
| |
| | Variable | Allowed dtypes | |
| | --- | --- | |
| | `T` | `float32`, `float16` | |
| |
| ## Files |
| |
| - [`metadata.json`](build/webgpu/metadata.json) — kernel metadata (id, digests, per-variant templates, 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 |
| - [`quick-gelu.wgsl.jinja`](build/webgpu/quick-gelu.wgsl.jinja) |
| |
| ## Use with `@huggingface/kernels` |
| |
| ```sh |
| npm install --save-exact @huggingface/kernels@0.0.1-preview.2 |
| ``` |
| |
| Required output shapes and logical data types are inferred from the supplied inputs and attributes; result tensors are allocated automatically. |
| |
| The `version: 1` option selects the published kernel contract; it is independent of any operator opset, contrib `since_version`, or model version. |
| It follows the `v1` branch as fixes land. To pin exact artifact bytes, pass a 40-character commit `revision` instead of `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/com.microsoft.QuickGelu", { version: 1 }); |
| const { Y } = await kernel({ X: { data: XData, shape: [2, 4] } }); |
| ``` |
|
|
