Struct LoraConfig 

pub struct LoraConfig {
    pub rank: usize,
    pub alpha: f64,
    pub dropout: f64,
    pub target_modules: Vec<String>,
    pub seed: u64,
}

Configuration for a LoRA adapter layer.

LoRA (Low-Rank Adaptation, Hu et al., 2021) decomposes weight updates as dW = B @ A where B in R^{d x r}, A in R^{r x k}, and r << min(d, k). The effective scaling factor applied is alpha / rank.

Fields

rank: usize

Low-rank dimension r.

alpha: f64

Scaling factor: effective scaling = alpha / rank.

dropout: f64

Dropout probability for training regularisation (0.0 = no dropout).

target_modules: Vec<String>

Which weight matrices to adapt (matched by name).

seed: u64

RNG seed for reproducible A-matrix initialisation.

Trait Implementations

impl Clone for LoraConfig

fn clone(&self) -> LoraConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for LoraConfig

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

Formats the value using the given formatter.

impl Default for LoraConfig

fn default() -> Self

Returns the “default value” for a type.