Enum TrainingMethod 

pub enum TrainingMethod {
    Supervised {
        batch_size: usize,
        epochs: usize,
    },
    RLHF {
        reward_weight: f32,
        kl_penalty: f32,
    },
    DPO {
        beta: f32,
        ref_weight: f32,
    },
    Online {
        lr_decay: f32,
        window_size: usize,
    },
    FewShot {
        k_shot: usize,
        meta_lr: f32,
    },
}

Training method configuration

Variants

Supervised

Standard supervised learning

Fields

batch_size: usize

Batch size for training

epochs: usize

Number of epochs

RLHF

Reinforcement learning from feedback

Fields

reward_weight: f32

Reward model weight

kl_penalty: f32

KL divergence penalty

DPO

Direct preference optimization

Fields

beta: f32

Beta parameter for DPO

ref_weight: f32

Reference model weight

Online

Continuous online learning

Fields

lr_decay: f32

Learning rate decay

window_size: usize

Window size for recent examples

FewShot

Few-shot adaptation

Fields

k_shot: usize

Number of examples per class

meta_lr: f32

Meta-learning rate

Trait Implementations

impl Clone for TrainingMethod

fn clone(&self) -> TrainingMethod

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for TrainingMethod

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

Formats the value using the given formatter.

impl Default for TrainingMethod

fn default() -> Self

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for TrainingMethod

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for TrainingMethod

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

where __S: Serializer,

Serialize this value into the given Serde serializer.