Forum
Login
Register

Welcome to our Support Center
UpdatedAugust 27, 2025

RotaryEmbedding

Description

RotaryEmbedding is the implementation of rotary positional embeddings (RoPE). The positions are represented as rotation matrices that are multiplied to query and key before the inner product of query and key is taken.

 

Input parameters

 

specified_outputs_name :Β array, this parameter lets you manually assign custom names to the output tensors of a node.

Β Graphs in :Β cluster, ONNX model architecture.

input (heterogeneous) – T : object, 3D tensor with shape (batch_size, sequence_length, hidden_size) or 4D with shape (batch_size, num_heads, sequence_length, head_size).
position_ids (heterogeneous) – M : object, 1D tensor with shape (1) or 2D tensor with shape (batch_size, sequence_length).
cos_cache (heterogeneous) – T : object, 2D tensor with shape (max_sequence_length, head_size / 2) or (max_sequence_length, rotary_embedding_dim / 2).
sin_cache (heterogeneous) – T : object, 2D tensor with shape (max_sequence_length, head_size / 2) or (max_sequence_length, rotary_embedding_dim / 2).

Β Parameters :Β cluster,

interleavedΒ :Β boolean, indicates whether the input has real and imaginary parts interleaved. If false, meaning the first half of the input consists of real values and the second half consists of imaginary values.
Default value β€œFalse”.
is_packed_batching :Β boolean, ragged batch inputs or not.
Default value β€œFalse”.
num_heads : integer, number of attention heads. Must use with rotary_embedding_dim.
Default value β€œ0”.
rotary_embedding_dim : integer, rotary embedding dimension.
Default value β€œ0”.
scale : float, custom scale will be used if specified.
Default value β€œ1”.
Β training?Β :Β boolean, whether the layer is in training mode (can store data for backward).
Default value β€œTrue”.
Β lda coeff :Β float, defines the coefficient by which the loss derivative will be multiplied before being sent to the previous layer (since during the backward run we go backwards).
Default value β€œ1”.

Β name (optional) :Β string,Β name of the node.

Output parameters

output (heterogeneous) – T : object, tensor with same shape as input.

Type Constraints

TΒ in (tensor(float),Β tensor(float16),Β tensor(bfloat16)) : Constrain input and output types to float tensors.

M in (tensor(int64)) : Constrain input and output types to integer tensors.

Example

All these exemples are snippets PNG, you can drop these Snippet onto the block diagram and get the depicted code added to your VI (Do not forget to install Deep Learning library to run it).
Table of Contents