Struct QLoraMapper 

pub struct QLoraMapper { /* private fields */ }

A module mapper implementing QLoRA: it quantizes the (frozen) base weights and attaches full-precision trainable LoRA adapters to 2-D weights.

The quantized base is kept at rest in its low-bit representation; the adapter contribution is added on top during the forward pass (the base is dequantized on the fly when composed).

Implementations

impl QLoraMapper

pub fn new(config: LoraConfig, quantizer: Quantizer) -> Self

Create a new QLoRA mapper from the LoRA configuration and a quantizer.

Trait Implementations

impl ModuleMapper for QLoraMapper

fn map_float<const D: usize>( &mut self, param: Param<Tensor<D>>, ) -> Param<Tensor<D>>

Map a float parameter in the module.

fn enter_module(&mut self, name: &str, container_type: &str)

Called when entering a submodule.

fn exit_module(&mut self, name: &str, container_type: &str)

Called when exiting a submodule.

fn map_int<const D: usize>( &mut self, param: Param<Tensor<D, Int>>, ) -> Param<Tensor<D, Int>>

Map an int parameter in the module.

fn map_bool<const D: usize>( &mut self, param: Param<Tensor<D, Bool>>, ) -> Param<Tensor<D, Bool>>

Map a bool parameter in the module.