| --- |
| library_name: kernels |
| license: apache-2.0 |
| tags: |
| - kernel |
| - webgpu |
| - wgsl |
| --- |
| # ai.onnx.BitCast |
|
|
| `ai.onnx` · standard ONNX operator · ONNX opset ≥ 26 |
|
|
| ## Description |
|
|
| Reinterprets the raw bit pattern of a tensor as a different data type without any value conversion. The target type must have the same bit-width as the input type, and the output tensor has the same shape as the input. |
|
|
| See the [ONNX `BitCast` spec](https://onnx.ai/onnx/operators/onnx__BitCast.html) for the reference semantics. |
|
|
| ## Inputs |
|
|
| | Name | Logical dtype | Rank | Shape | Description | Presence | |
| | --- | --- | --- | --- | --- | --- | |
| | `input` | `T` | — | — | Input tensor to be bitwise reinterpreted. | required | |
|
|
| ## Outputs |
|
|
| | Name | Logical dtype | Rank | Shape | Description | Presence | |
| | --- | --- | --- | --- | --- | --- | |
| | `output` | `U` | same as `input` | same as `input` | Output tensor with the same shape as the input, reinterpreted as the target type. | required | |
|
|
| ## Attributes |
|
|
| Attributes and default values (overridable per request): |
|
|
| | Attribute | Default | Description | |
| | --- | --- | --- | |
| | `to` | — | Required TensorProto DataType enum integer naming the output dtype; the target type must have the same bit-width as the input type. | |
|
|
| ## Type constraints |
|
|
| | Variable | Allowed dtypes | |
| | --- | --- | |
| | `T` | `float32`, `int8`, `int32`, `uint8`, `uint32` | |
| | `U` | `float32`, `int8`, `int32`, `uint8`, `uint32` | |
|
|
| ## 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 cases |
| - [`bitcast.wgsl.jinja`](build/webgpu/bitcast.wgsl.jinja) |
|
|
| ## Use with `@huggingface/kernels` |
|
|
| ```sh |
| npm install --save-exact @huggingface/kernels@0.0.1-preview.3 |
| ``` |
|
|
| 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/ai.onnx.BitCast", { version: 1 }); |
| const { output } = await kernel({ input: { data: inputData, shape: [3] } }, { |
| attrs: { to: 1 }, |
| }); |
| ``` |
|
|
