| --- |
| library_name: kernels |
| license: apache-2.0 |
| tags: |
| - kernel |
| - webgpu |
| - wgsl |
| --- |
| # ai.onnx.LpNormalization |
|
|
| `ai.onnx` · standard ONNX operator · ONNX opset ≥ 1 |
|
|
| ## Description |
|
|
| Applies Lp-normalization to the input tensor along the specified axis: `output = input / Lp_norm(input, axis)`. Supports L1 (`p=1`) and L2 (`p=2`) norms. Where the Lp norm is zero, the output is defined as zero to avoid division by zero. |
|
|
| See the [ONNX `LpNormalization` spec](https://onnx.ai/onnx/operators/onnx__LpNormalization.html) for the reference semantics. |
|
|
| ## Inputs |
|
|
| | Name | Bind key | Logical dtype | Rank | Shape | Description | Presence | |
| | --- | --- | --- | --- | --- | --- | --- | |
| | `input` | `input` | `T` | — | — | Input tensor to normalize. | required | |
|
|
| ## Outputs |
|
|
| | Name | Bind key | Logical dtype | Rank | Shape | Description | Presence | |
| | --- | --- | --- | --- | --- | --- | --- | |
| | `output` | `output` | `T` | same as `input` | same as `input` | Tensor after Lp-normalization; same shape as the input. | required | |
|
|
| ## Attributes |
|
|
| Default values (overridable per request): |
|
|
| | Attribute | Default | Description | |
| | --- | --- | --- | |
| | `axis` | `-1` | The axis along which normalization is applied; `-1` means the last axis. | |
| | `p` | `2` | The order of the Lp norm to use; only `1` (L1) or `2` (L2) are supported. | |
|
|
| ## Type constraints |
|
|
| | Variable | Allowed dtypes | |
| | --- | --- | |
| | `T` | `float32`, `float16` | |
|
|
| ## 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 |
| - [`lp-norm-divide.wgsl.jinja`](build/webgpu/lp-norm-divide.wgsl.jinja) |
| - [`lp-norm-reduce.wgsl.jinja`](build/webgpu/lp-norm-reduce.wgsl.jinja) |
| - [`lp-norm-split-combine.wgsl.jinja`](build/webgpu/lp-norm-split-combine.wgsl.jinja) |
| - [`norm-row-stats.wgsl.jinja`](build/webgpu/norm-row-stats.wgsl.jinja) |
|
|
| ## Use with `@huggingface/kernels` |
|
|
| The loader derives every required output's shape and logical dtype from the manifest contract and this call. |
| It then allocates the result tensors automatically. |
|
|
| 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.LpNormalization", { version: 1 }); |
| const { output } = await kernel({ input: { data: inputData, shape: [2, 2] } }); |
| ``` |
|
|
