ai.onnx.GRU

ai.onnx · standard ONNX operator · ONNX opset ≥ 22

Description

Computes a single-layer Gated Recurrent Unit (GRU) over an input sequence, applying update (z), reset (r), and hidden (h) gates at each time step. Supports forward, reverse, and bidirectional traversal; activation functions default to Sigmoid for f and Tanh for g.

See the ONNX GRU spec for the reference semantics.

Inputs

Name	Upstream name	Logical dtype	Rank	Shape	Description	Presence
`x`	`X`	`T`	`3`	—	Input sequences with shape `[seq_length, batch_size, input_size]` when `layout=0`, or `[batch_size, seq_length, input_size]` when `layout=1`.	required
`w`	`W`	`T`	`3`	—	Gate weight tensor with shape `[num_directions, 3*hidden_size, input_size]`, ordered as `W[zrh]` within each direction.	required
`r`	`R`	`T`	`3`	—	Recurrence weight tensor with shape `[num_directions, 3*hidden_size, hidden_size]`, ordered as `R[zrh]` within each direction.	required
`b`	`B`	`T`	`2`	—	Bias tensor for the gates, shape `[num_directions, 6*hidden_size]`, concatenating input and recurrence biases. ONNX defines an omitted bias as zero; this package requires an explicit tensor, which may be zero-filled to represent omission.	required
`sequence_lens`	—	`int32`	`1`	—	Per-batch sequence lengths of shape `[batch_size]`; assumed all equal to `seq_length` if absent.	optional
`initial_h`	—	`T`	`3`	—	Initial hidden state with shape `[num_directions, batch_size, hidden_size]` when `layout=0`, or `[batch_size, num_directions, hidden_size]` when `layout=1`; assumed zero if absent.	optional

Outputs

Name	Upstream name	Logical dtype	Rank	Shape	Description	Presence
`y`	`Y`	`T`	`4`	derived	Intermediate hidden outputs for all time steps, with shape `[seq_length, num_directions, batch_size, hidden_size]` when `layout=0`, or `[batch_size, seq_length, num_directions, hidden_size]` when `layout=1`.	required
`y_h`	`Y_h`	`T`	`3`	derived	Final hidden state with shape `[num_directions, batch_size, hidden_size]` when `layout=0`, or `[batch_size, num_directions, hidden_size]` when `layout=1`.	required

Attributes

Attributes and default values (overridable per request):

Attribute	Default	Description
`activation_alpha`	—	Optional alpha parameters for activation functions that use alpha, consumed in activation-list order; omitted entries use the ONNX defaults for their activation.
`activation_beta`	—	Optional beta parameters for activation functions that use beta, consumed in activation-list order; omitted entries use the ONNX defaults for their activation.
`activations`	—	Activation functions for the update/reset gates and candidate state. Defaults to `["Sigmoid", "Tanh"]` per direction.
`clip`	—	Optional non-negative threshold applied to activation inputs as `[-clip, +clip]`; omission disables clipping, while an explicit `0` clamps them to zero.
`direction`	`"forward"`	Specifies traversal direction: `"forward"` (default), `"reverse"`, or `"bidirectional"`.
`hidden_size`	—	Optional number of neurons in the hidden layer; when omitted, it is inferred from the W and R tensor shapes.
`layout`	`0`	Controls the shape format of `X`, `initial_h`, `Y`, and `Y_h`: `0` uses `[seq_length, batch_size, ...]` ordering (default), `1` uses `[batch_size, seq_length, ...]`.
`linear_before_reset`	`0`	When non-zero, applies the recurrence linear transform to the hidden state before multiplying by the reset gate output.

Type constraints

Variable	Allowed dtypes
`T`	`float32`

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
datamove-transpose-2d-tiled.wgsl.jinja
dense-tiled-matmul.wgsl.jinja
gru-general.wgsl.jinja
gru-parallel-global.wgsl.jinja
gru-recur-candidate-step.wgsl.jinja
gru-recur-candidate-subgroup-step.wgsl.jinja
gru-recur-gates-step.wgsl.jinja
gru-recur-gates-subgroup-step.wgsl.jinja
gru-recur-init-step.wgsl.jinja
gru-timestep-body.wgsl.jinja
rnn-input-proj.wgsl.jinja
rnn-sequence-mask.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.GRU", { version: 1 });
const { y, y_h } = await kernel({
  x: { data: xData, shape: [1, 1, 1] },
  w: { data: wData, shape: [1, 3, 1] },
  r: { data: rData, shape: [1, 3, 1] },
  b: { data: bData, shape: [1, 6] },
});

Downloads last month: -

kernel

webgpu

wgsl

apache-2.0

WebGPU

Requires WebGPU support. See the compatibility table.