Struct MultiHeadAttention 

pub struct MultiHeadAttention {
    pub query: Linear,
    pub key: Linear,
    pub value: Linear,
    pub output: Linear,
    pub dropout: Dropout,
    pub activation: Gelu,
    pub d_model: usize,
    pub n_heads: usize,
    pub d_k: usize,
    pub min_float: f64,
    pub quiet_softmax: bool,
}

The multihead attention module as describe in the paper Attention Is All You Need.

Params

• query: Linear layer with d_model input and output features.
• key: Linear layer with d_model input and output features.
• value: Linear layer with d_model input and output features.
• output: Linear layer with d_model input and output features.

Should be created with MultiHeadAttentionConfig.

Fields

query: Linear

Linear layer to transform the input features into the query space.

key: Linear

Linear layer to transform the input features into the key space.

value: Linear

Linear layer to transform the input features into the value space.

output: Linear

Linear layer to transform the output features back to the original space.

dropout: Dropout

Dropout layer.

activation: Gelu

Activation function.

d_model: usize

The size of each linear layer.

n_heads: usize

The number of heads.

d_k: usize

Size of the key and query vectors.

min_float: f64

Minimum value a float can take.

quiet_softmax: bool

Use “quiet softmax” instead of regular softmax.

Implementations

impl MultiHeadAttention

pub fn forward(&self, input: MhaInput) -> MhaOutput

Applies the forward pass on the input tensors.

See MultiHeadAttention for more information.

Shapes

• query: [batch_size, seq_length_1, d_model]
• key: [batch_size, seq_length_2, d_model]
• value: [batch_size, seq_length_2, d_model]
• output: [batch_size, seq_length_1, d_model]

Panics

Panics if the last axis of query, key or value is not d_model, if the batch size of key or value differs from query, or if the sequence length of value differs from key.

pub fn forward_cache(&self, input: MhaInput, cache: &mut MhaCache) -> MhaOutput

Applies the forward pass using a cache.

Shapes

• query: [batch_size, seq_length_1, d_model]
• key: [batch_size, seq_length_2, d_model]
• value: [batch_size, seq_length_2, d_model]
• output: [batch_size, seq_length_1, d_model]

Panics

Panics if the last axis of query, key or value is not d_model, if the batch size of key or value differs from query, or if the sequence length of value differs from key.

Trait Implementations

impl Clone for MultiHeadAttention

fn clone(&self) -> MultiHeadAttention

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for MultiHeadAttention

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

Formats the value using the given formatter.

impl Display for MultiHeadAttention

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

Formats the value using the given formatter.

impl Module for MultiHeadAttention

fn num_params(&self) -> usize

Get the number of parameters the module has, including all of its sub-modules, without initializing the ones not initialized yet.

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

where Visitor: ModuleVisitor,

Visit each parameter and module-owned control value with a visitor.

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

where Mapper: ModuleMapper,

Map each parameter and module-owned control value with a mapper.

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

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

fn to_device(self, device: &Device) -> MultiHeadAttention

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

fn fork(self, device: &Device) -> MultiHeadAttention

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

fn valid(&self) -> MultiHeadAttention

Returns a validation snapshot without autodiff or active training flags.

fn train(self) -> MultiHeadAttention

Enables autodiff and restores configured training state throughout the module.

fn materialize(self) -> MultiHeadAttention

Fold parameter reparameterizations into their effective values and remove them.

fn devices(&self) -> Vec<Device>

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

fn set_require_grad(self, require_grad: bool) -> Self

Set whether every floating-point tensor parameter in the module tree requires gradients.

fn set_require_grad_group(self, group: ParamGroup, require_grad: bool) -> Self

Set whether matched floating-point tensor parameters require gradients.

fn no_grad(self) -> Self

Disable gradients for every floating-point tensor parameter in the module tree.

fn freeze(self) -> Self

Freeze the whole module tree.

fn unfreeze(self) -> Self

Unfreeze the whole module tree.

fn freeze_group(self, group: ParamGroup) -> Self

Freeze every module-owned value in the given group, leaving the rest of the module untouched.

fn unfreeze_group(self, group: ParamGroup) -> Self

Unfreeze every module-owned value in the given group, leaving the rest of the module untouched.

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

Quantize the weights of the module.

fn quantize_weights_group( self, quantizer: &mut Quantizer, group: ParamGroup, ) -> Self

Quantize the weights of the given parameter group.

fn apply_reparameterization<R>(self, reparameterizer: R) -> Self

where Self: Sized, R: Reparameterizer,

Attach reparameterizations using the given Reparameterizer.

fn apply_lora(self, lora: Lora) -> Self

where Self: Sized,

Attach LoRA adapters to the module’s 2-D weights, freezing all base tensor parameters.

fn apply_qlora(self, qlora: QLora) -> Self

where Self: Sized,

Apply QLoRA to the module: quantize the (frozen) base tensor parameters and attach trainable LoRA adapters to 2-D weights.

fn into_record(self) -> ModuleRecord

where Self: Sized,

Collect this module’s tensor parameters into a ModuleRecord.

fn into_record_group(self, group: ParamGroup) -> ModuleRecord

where Self: Sized,

Collect the tensor parameters group names into a ModuleRecord.

fn try_load_record(self, record: ModuleRecord) -> Result<Self, RecordError>

where Self: Sized,

Apply a ModuleRecord to this module, returning the loaded module.

fn load_record(self, record: ModuleRecord) -> Self

where Self: Sized,

Apply a ModuleRecord to this module, consuming and returning it.

fn save_file<P>(self, path: P) -> Result<(), RecordError>

where P: AsRef<Path>, Self: Sized,

Save this module’s parameters to a burnpack file on disk.

fn load_file<P>(self, path: P) -> Self

where P: AsRef<Path>, Self: Sized,

Load this module’s parameters from a burnpack file on disk, returning the loaded module.

fn try_load_file<P>(self, path: P) -> Result<Self, RecordError>

where P: AsRef<Path>, Self: Sized,

Fallible variant of load_file.

impl ModuleDisplay for MultiHeadAttention

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 ModuleDisplayDefault for MultiHeadAttention

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.