Struct RuntimeConfig 

pub struct RuntimeConfig {
    pub batch_size: u32,
    pub beam_width: u32,
    pub isl: u32,
    pub osl: u32,
    pub prefix: u32,
    pub seq_imbalance_correction_scale: f64,
    pub gen_seq_imbalance_correction_scale: f64,
}

Per-call runtime inputs. Field-for-field mirror of the Python sdk/config.RuntimeConfig.

The imbalance-correction scales thread into the per-op queries exactly where Python applies them (base_backend.py:331,372): context-attention ops multiply by seq_imbalance_correction_scale, generation-attention ops by gen_seq_imbalance_correction_scale. (The FPM telemetry path has no scale concept and keeps 1.0.)

Fields

batch_size: u32

beam_width: u32

Beam width. The generation phase queries token-major ops at x = batch_size * beam_width (Python _run_generation_phase); attention ops key on the raw decode batch.

isl: u32

osl: u32

prefix: u32

Cached tokens already in the KV cache (context phase only).

seq_imbalance_correction_scale: f64

Context-attention sequence-imbalance correction (default 1.0).

gen_seq_imbalance_correction_scale: f64

Generation-attention sequence-imbalance correction (default 1.0).

Trait Implementations

impl Clone for RuntimeConfig

fn clone(&self) -> RuntimeConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Copy for RuntimeConfig

impl Debug for RuntimeConfig

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

Formats the value using the given formatter.

impl Default for RuntimeConfig

fn default() -> Self

Returns the “default value” for a type.

impl PartialEq for RuntimeConfig

fn eq(&self, other: &RuntimeConfig) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for RuntimeConfig