Skip to main content

TransformerDecoderLayerParams

Struct TransformerDecoderLayerParams 

Source
pub struct TransformerDecoderLayerParams<'a, T: FloatElement + AttentionScalar + RealScalar, B: BackendOps<T> + AttentionOps<T> + Default> {
Show 20 fields pub norm1_weight: &'a Var<T, B>, pub norm1_bias: &'a Var<T, B>, pub self_attn: MhaProjectionParams<'a, T, B>, pub norm2_weight: &'a Var<T, B>, pub norm2_bias: &'a Var<T, B>, pub cross_attn: MhaProjectionParams<'a, T, B>, pub norm3_weight: &'a Var<T, B>, pub norm3_bias: &'a Var<T, B>, pub ffn_w1: &'a Var<T, B>, pub ffn_b1: Option<&'a Var<T, B>>, pub ffn_w2: &'a Var<T, B>, pub ffn_b2: Option<&'a Var<T, B>>, pub self_attn_residual_dropout_p: f64, pub self_attn_residual_training: bool, pub cross_attn_residual_dropout_p: f64, pub cross_attn_residual_training: bool, pub ffn_hidden_dropout_p: f64, pub ffn_hidden_training: bool, pub ffn_residual_dropout_p: f64, pub ffn_residual_training: bool,
}
Expand description

Borrowed parameters for functional transformer decoder-layer execution.

Fields§

§norm1_weight: &'a Var<T, B>

LayerNorm1 gamma [d_model].

§norm1_bias: &'a Var<T, B>

LayerNorm1 beta [d_model].

§self_attn: MhaProjectionParams<'a, T, B>

Masked self-attention projections.

§norm2_weight: &'a Var<T, B>

LayerNorm2 gamma [d_model].

§norm2_bias: &'a Var<T, B>

LayerNorm2 beta [d_model].

§cross_attn: MhaProjectionParams<'a, T, B>

Cross-attention projections.

§norm3_weight: &'a Var<T, B>

LayerNorm3 gamma [d_model].

§norm3_bias: &'a Var<T, B>

LayerNorm3 beta [d_model].

§ffn_w1: &'a Var<T, B>

FFN linear1 weight [d_ff, d_model].

§ffn_b1: Option<&'a Var<T, B>>

FFN linear1 bias [d_ff].

§ffn_w2: &'a Var<T, B>

FFN linear2 weight [d_model, d_ff].

§ffn_b2: Option<&'a Var<T, B>>

FFN linear2 bias [d_model].

§self_attn_residual_dropout_p: f64

Dropout probability after masked self-attention.

§self_attn_residual_training: bool

Training mode for self-attention residual dropout.

§cross_attn_residual_dropout_p: f64

Dropout probability after cross-attention.

§cross_attn_residual_training: bool

Training mode for cross-attention residual dropout.

§ffn_hidden_dropout_p: f64

Dropout probability inside FFN hidden projection.

§ffn_hidden_training: bool

Training mode for FFN hidden dropout.

§ffn_residual_dropout_p: f64

Dropout probability after FFN residual branch.

§ffn_residual_training: bool

Training mode for FFN residual dropout.

Auto Trait Implementations§

§

impl<'a, T, B> !RefUnwindSafe for TransformerDecoderLayerParams<'a, T, B>

§

impl<'a, T, B> !UnwindSafe for TransformerDecoderLayerParams<'a, T, B>

§

impl<'a, T, B> Forget for TransformerDecoderLayerParams<'a, T, B>

§

impl<'a, T, B> Freeze for TransformerDecoderLayerParams<'a, T, B>
where &'a Var<T, B>: Freeze, MhaProjectionParams<'a, T, B>: Freeze, Option<&'a Var<T, B>>: Freeze,

§

impl<'a, T, B> Move for TransformerDecoderLayerParams<'a, T, B>

§

impl<'a, T, B> Send for TransformerDecoderLayerParams<'a, T, B>
where &'a Var<T, B>: Send, MhaProjectionParams<'a, T, B>: Send, Option<&'a Var<T, B>>: Send,

§

impl<'a, T, B> Sync for TransformerDecoderLayerParams<'a, T, B>
where &'a Var<T, B>: Sync, MhaProjectionParams<'a, T, B>: Sync, Option<&'a Var<T, B>>: Sync,

§

impl<'a, T, B> Unpin for TransformerDecoderLayerParams<'a, T, B>
where &'a Var<T, B>: Unpin, MhaProjectionParams<'a, T, B>: Unpin, Option<&'a Var<T, B>>: Unpin,

§

impl<'a, T, B> UnsafeUnpin for TransformerDecoderLayerParams<'a, T, B>
where &'a Var<T, B>: UnsafeUnpin, MhaProjectionParams<'a, T, B>: UnsafeUnpin, Option<&'a Var<T, B>>: UnsafeUnpin,

Blanket Implementations§

Source§

impl<T> Any for T
where T: 'static + ?Sized,

Source§

fn type_id(&self) -> TypeId

Gets the TypeId of self. Read more
Source§

impl<T> ArchivePointee for T

Source§

type ArchivedMetadata = ()

The archived version of the pointer metadata for this type.
Source§

fn pointer_metadata( _: &<T as ArchivePointee>::ArchivedMetadata, ) -> <T as Pointee>::Metadata

Converts some archived metadata to the pointer metadata for itself.
Source§

impl<T> Borrow<T> for T
where T: ?Sized,

Source§

fn borrow(&self) -> &T

Immutably borrows from an owned value. Read more
Source§

impl<T> BorrowMut<T> for T
where T: ?Sized,

Source§

fn borrow_mut(&mut self) -> &mut T

Mutably borrows from an owned value. Read more
Source§

impl<T> From<T> for T

Source§

fn from(t: T) -> T

Returns the argument unchanged.

Source§

impl<T> Instrument for T

Source§

fn instrument(self, span: Span) -> Instrumented<Self> ⓘ

Instruments this type with the provided Span, returning an Instrumented wrapper. Read more
Source§

fn in_current_span(self) -> Instrumented<Self> ⓘ

Instruments this type with the current Span, returning an Instrumented wrapper. Read more
Source§

impl<T, U> Into<U> for T
where U: From<T>,

Source§

fn into(self) -> U

Calls U::from(self).

That is, this conversion is whatever the implementation of From<T> for U chooses to do.

Source§

impl<T> LayoutRaw for T

Source§

fn layout_raw(_: <T as Pointee>::Metadata) -> Result<Layout, LayoutError>

Returns the layout of the type.
Source§

impl<T, N1, N2> Niching<NichedOption<T, N1>> for N2
where T: SharedNiching<N1, N2>, N1: Niching<T>, N2: Niching<T>,

Source§

unsafe fn is_niched(niched: *const NichedOption<T, N1>) -> bool

Returns whether the given value has been niched. Read more
Source§

fn resolve_niched(out: Place<NichedOption<T, N1>>)

Writes data to out indicating that a T is niched.
Source§

impl<T> Pointee for T

Source§

type Metadata = ()

The metadata type for pointers and references to this type.
Source§

impl<T> Same for T

Source§

type Output = T

Should always be Self
Source§

impl<T, U> TryFrom<U> for T
where U: Into<T>,

Source§

type Error = !

The type returned in the event of a conversion error.
Source§

fn try_from(value: U) -> Result<T, !>

Performs the conversion.
Source§

impl<T, U> TryInto<U> for T
where U: TryFrom<T>,

Source§

type Error = <U as TryFrom<T>>::Error

The type returned in the event of a conversion error.
Source§

fn try_into(self) -> Result<U, <U as TryFrom<T>>::Error>

Performs the conversion.
Source§

impl<T> WithSubscriber for T

Source§

fn with_subscriber<S>(self, subscriber: S) -> WithDispatch<Self> ⓘ
where S: Into<Dispatch>,

Attaches the provided Subscriber to this type, returning a WithDispatch wrapper. Read more
Source§

fn with_current_subscriber(self) -> WithDispatch<Self> ⓘ

Attaches the current default Subscriber to this type, returning a WithDispatch wrapper. Read more