Struct QLoraTrainingConfig 

pub struct QLoraTrainingConfig {
    pub adapter_config: AdapterTrainingConfig,
    pub num_epochs: usize,
    pub batch_size: usize,
    pub log_every: usize,
    pub save_every: Option<usize>,
    pub warmup_steps: usize,
    pub use_paged_optimizer: bool,
    pub page_size: usize,
    pub max_optimizer_memory: usize,
}

Configuration for QLoRA training.

Fields

adapter_config: AdapterTrainingConfig

Adapter training configuration (from peft-rs).

num_epochs: usize

Number of training epochs.

batch_size: usize

Batch size for training.

log_every: usize

Logging frequency (steps).

save_every: Option<usize>

Checkpoint save frequency (steps, None = no checkpoints).

warmup_steps: usize

Warmup steps for learning rate.

use_paged_optimizer: bool

Use paged optimizer (CPU offload for optimizer states).

page_size: usize

Page size for paged optimizer (bytes).

max_optimizer_memory: usize

Maximum memory for optimizer states on GPU (bytes, 0 = unlimited).

Trait Implementations

impl Clone for QLoraTrainingConfig

fn clone(&self) -> QLoraTrainingConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for QLoraTrainingConfig

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

Formats the value using the given formatter.

impl Default for QLoraTrainingConfig

fn default() -> QLoraTrainingConfig

Returns the “default value” for a type.