Struct BaseLoRA 

pub struct BaseLoRA {
    pub layers: Vec<LoRALayer>,
    pub rank: usize,
    pub hidden_dim: usize,
    pub alpha: f32,
}

Base LoRA for background adaptation

Higher rank (4-16) for more expressive adaptation. Applied hourly during background learning cycles.

Fields

layers: Vec<LoRALayer>

LoRA layers

rank: usize

Rank

hidden_dim: usize

Hidden dimension

alpha: f32

Alpha scaling factor

Implementations

impl BaseLoRA

pub fn new(hidden_dim: usize, rank: usize, num_layers: usize) -> Self

Create new Base LoRA

pub fn forward_layer(&self, layer_idx: usize, input: &[f32], output: &mut [f32])

Forward pass for single layer

pub fn merge_into(&self, model_weights: &mut [f32], layer_idx: usize)

Merge LoRA weights into model weights (for inference optimization)

pub fn num_layers(&self) -> usize

Get number of layers

pub fn param_count(&self) -> usize

Get total parameter count

pub fn get_layer_weights( &self, layer_idx: usize, ) -> Option<(&Vec<f32>, &Vec<f32>)>

Get weights for a specific layer for export (lora_a, lora_b)

Trait Implementations

impl Clone for BaseLoRA

fn clone(&self) -> BaseLoRA

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for BaseLoRA

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for BaseLoRA

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for BaseLoRA

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.