Struct RotaryEncoding

pub struct RotaryEncoding<B>
where
    B: Backend,
{
    pub freq_complex: Tensor<B, 3>,
    pub theta: Tensor<B, 1>,
    /* private fields */
}

A module that applies rotary positional encoding to a tensor. Rotary Position Encoding or Embedding (RoPE), is a type of position embedding which encodes absolute positional information with rotation matrix and naturally incorporates explicit relative position dependency in self-attention formulation.

Introduced in the paper: RoFormer: Enhanced Transformer with Rotary Position Embedding

Should be created using RotaryEncodingConfig.

Fields

freq_complex: Tensor<B, 3>

Complex frequency tensor of shape (max_sequence_length, d_model, 2) with real and imaginary components

theta: Tensor<B, 1>

Frequency vector used to compute/apply the complex rotations.

Implementations

impl<B> RotaryEncoding<B>

where B: Backend,

pub fn forward<const D: usize>(&self, x: Tensor<B, D>) -> Tensor<B, D>

Applies rotary positional encoding to a tensor of dimensions (…, seq_len, d_model)

Arguments:

• x - Input tensor of shape (…, seq_len, d_model). Accommodate both 3D and 4D tensors for (batch size, seq_len, hidden_dim) or (batch size, num_heads, seq_len, hidden_dim) respectively.

Returns:

Output tensor with the same shape as input tensor after applying rotary encoding.

Panics

If the input tensor does not have at least 2 dimensions for sequence length and hidden dimension.

pub fn apply<const D: usize>( &self, x: Tensor<B, D>, start: usize, ) -> Tensor<B, D>

Applies rotary positional encoding to a tensor of dimensions (…, seq_len, d_model)

Arguments:

• x - Input tensor of shape (…, seq_len, d_model). Accommodate both 3D and 4D tensors for (batch size, seq_len, hidden_dim) or (batch size, num_heads, seq_len, hidden_dim) respectively.
• start - Sequence start position index.

Returns:

Output tensor with the same shape as input tensor after applying rotary encoding.

Panics

If the input tensor does not have at least 2 dimensions for sequence length and hidden dimension.

pub fn shift(&mut self, start: usize)

Shifts the pre-computed rotary frequency to cover a new range of positions.

This method updates the internal frequency tensor freq_complex to store the rotary positional encodings for a new window of positions starting at start.

Trait Implementations

impl<B> AutodiffModule<B> for RotaryEncoding<B>

where B: Backend + AutodiffBackend, <B as AutodiffBackend>::InnerBackend: Backend,

type InnerModule = RotaryEncoding<<B as AutodiffBackend>::InnerBackend>

Inner module without auto-differentiation.

fn valid(&self) -> <RotaryEncoding<B> as AutodiffModule<B>>::InnerModule

Get the same module, but on the inner backend without auto-differentiation.

impl<B> Clone for RotaryEncoding<B>

where B: Backend,

fn clone(&self) -> RotaryEncoding<B>

Returns a copy of the value. Read more

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source. Read more

impl<B> Debug for RotaryEncoding<B>

where B: Debug + Backend,

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter. Read more

impl<B> Display for RotaryEncoding<B>

where B: Backend,

fn fmt(&self, f: &mut Formatter<'_>) -> Result<(), Error>

Formats the value using the given formatter. Read more

impl<B> Module<B> for RotaryEncoding<B>

where B: Backend,

type Record = RotaryEncodingRecord<B>

Type to save and load the module.

fn load_record( self, record: <RotaryEncoding<B> as Module<B>>::Record, ) -> RotaryEncoding<B>

Load the module state from a record.

fn into_record(self) -> <RotaryEncoding<B> as Module<B>>::Record

Convert the module into a record containing the state.

fn num_params(&self) -> usize

Get the number of parameters the module has, including all of its sub-modules.

fn visit<Visitor>(&self, visitor: &mut Visitor)

where Visitor: ModuleVisitor<B>,

Visit each tensor parameter in the module with a visitor.

fn map<Mapper>(self, mapper: &mut Mapper) -> RotaryEncoding<B>

where Mapper: ModuleMapper<B>,

Map each tensor parameter in the module with a mapper.

fn collect_devices( &self, devices: Vec<<B as Backend>::Device>, ) -> Vec<<B as Backend>::Device>

Return all the devices found in the underneath module tree added to the given vector without duplicates.

fn to_device(self, device: &<B as Backend>::Device) -> RotaryEncoding<B>

Move the module and all of its sub-modules to the given device.

fn fork(self, device: &<B as Backend>::Device) -> RotaryEncoding<B>

Fork the module and all of its sub-modules to the given device.

fn devices(&self) -> Vec<<B as Backend>::Device>

Return all the devices found in the underneath module tree without duplicates.

fn no_grad(self) -> Self

Each tensor in the module tree will not require grad.

fn save_file<FR, PB>( self, file_path: PB, recorder: &FR, ) -> Result<(), RecorderError>

where FR: FileRecorder<B>, PB: Into<PathBuf>,

Save the module to a file using the provided file recorder.

fn load_file<FR, PB>( self, file_path: PB, recorder: &FR, device: &<B as Backend>::Device, ) -> Result<Self, RecorderError>

where FR: FileRecorder<B>, PB: Into<PathBuf>,

Load the module from a file using the provided file recorder.

fn quantize_weights(self, quantizer: &mut Quantizer) -> Self

Quantize the weights of the module.

impl<B> ModuleDisplay for RotaryEncoding<B>

where B: Backend,

fn custom_settings(&self) -> Option<DisplaySettings>

Custom display settings for the module.

fn custom_content(&self, content: Content) -> Option<Content>

Custom attributes for the module.

fn format(&self, passed_settings: DisplaySettings) -> String

Formats the module with provided display settings.

impl<B> ModuleDisplayDefault for RotaryEncoding<B>

where B: Backend,

fn content(&self, content: Content) -> Option<Content>

Attributes of the module used for display purposes.

fn num_params(&self) -> usize

Gets the number of the parameters of the module.