Struct Linear 

pub struct Linear<B>
where
    B: Backend,
{ /* private fields */ }

A fully-connected (dense) layer: y = xW^T + b.

Type Parameters

• B: the compute backend

Examples

let linear = Linear::<CpuBackend>::new(784, 128, true, DType::F32, &dev)?;
let x = CpuTensor::rand((32, 784), DType::F32, &dev)?; // batch of 32
let y = linear.forward(&x)?; // shape: [32, 128]

Implementations

impl<B> Linear<B>

where B: Backend,

pub fn new( in_features: usize, out_features: usize, use_bias: bool, dtype: DType, device: &<B as Backend>::Device, ) -> Result<Linear<B>, Error>

Create a new Linear layer with Kaiming uniform initialization.

Arguments

• in_features: size of each input sample
• out_features: size of each output sample
• use_bias: whether to add a learnable bias
• dtype: data type for parameters
• device: device to create parameters on

pub fn from_tensors( weight: Tensor<B>, bias: Option<Tensor<B>>, ) -> Result<Linear<B>, Error>

Create a Linear layer from existing weight and bias tensors. Useful for loading pre-trained models.

pub fn in_features(&self) -> usize

The input feature dimension.

pub fn out_features(&self) -> usize

The output feature dimension.

pub fn weight(&self) -> &Tensor<B>

Direct access to the weight tensor.

pub fn bias(&self) -> Option<&Tensor<B>>

Direct access to the bias tensor (if any).

Trait Implementations

impl<B> Module<B> for Linear<B>

where B: Backend,

fn forward(&self, x: &Tensor<B>) -> Result<Tensor<B>, Error>

Forward pass: y = x @ W^T + b

Input shape: [batch, in_features] Output shape: [batch, out_features]

fn parameters(&self) -> Vec<Tensor<B>>

Return all trainable parameters of this module. The optimizer uses these to update weights during training.

fn named_parameters(&self) -> Vec<(String, Tensor<B>)>

Return all trainable parameters with human-readable names.

fn set_training(&self, _training: bool)

Set training or evaluation mode.

fn is_training(&self) -> bool

Whether the module is in training mode (default: true).

fn train(&self)

Convenience: set training mode.

fn eval(&self)

Convenience: set evaluation mode.

fn num_parameters(&self) -> usize

Total number of scalar parameters in this module.

fn trainable_params_count(&self) -> usize

Number of trainable (variable) parameters.

fn frozen_parameters(&self) -> Vec<Tensor<B>>

Freeze all parameters: returns new parameter tensors with is_variable = false, preventing gradient accumulation.

fn state_dict(&self) -> Vec<(String, Tensor<B>)>

Returns a state_dict-style map of parameter name → tensor.