ai.onnx.SwiGLU

ai.onnx · standard ONNX operator · ONNX opset ≥ 28

Description

Applies the Swish activation to gate input A and multiplies it elementwise by linear input B: Y = (A * sigmoid(alpha * A)) * B. Inputs must have identical shapes; broadcasting is not part of this operator. This package supports float16 and float32; standard bfloat16 and double tensors are unsupported.

See the ONNX SwiGLU spec for the reference semantics.

Inputs

Outputs

Attributes

Default values (overridable per request):

Type constraints

Files

• metadata.json — kernel metadata (id, digests, per-variant templates, provenance)
• manifest.json — the op contract (source of truth)
• test.json — correctness cases
• bench.json — benchmark + tuning cases
• swiglu-elementwise.wgsl.jinja

Use with @huggingface/kernels

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.

import { getKernel } from "@huggingface/kernels";

const kernel = await getKernel("webgpu-kernels/ai.onnx.SwiGLU", { version: 1 });
const { y } = await kernel({ a: { data: aData, shape: [2, 4] }, b: { data: bData, shape: [2, 4] } });