Struct GreedyRepetitionPenalty 

pub struct GreedyRepetitionPenalty { /* private fields */ }

Sparse repetition-penalty metadata for model-side greedy argmax.

The token list is request-local and de-duplicated. Applying the penalty before GPU argmax avoids downloading full [batch, vocab] logits for the common greedy chat path while preserving repeat avoidance.

Implementations

impl GreedyRepetitionPenalty

pub fn new(penalty: f32, token_ids: Vec<u32>) -> Self

pub const fn penalty(&self) -> f32

pub fn token_ids(&self) -> &[u32]

pub fn is_empty(&self) -> bool

Trait Implementations

impl Clone for GreedyRepetitionPenalty

fn clone(&self) -> GreedyRepetitionPenalty

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for GreedyRepetitionPenalty

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

Formats the value using the given formatter.