Struct ModelSamplingContext 

pub struct ModelSamplingContext {
    pub is_reasoning: bool,
    pub defaults_origin: Option<DefaultsOrigin>,
}

Everything about the target model that changes how sampling resolves, independent of any specific request.

Bundled rather than passed as separate parameters because both InferenceConfig::resolve_with_profile and crate::request_pipeline::sampling::resolve_sampling need the same two facts about the same model, and the list has already grown once (see #685) — a named struct reads at call sites instead of two easily transposed booleans.

Fields

is_reasoning: bool

Whether the model carries gglib’s reasoning capability tag. Selects the coupled-trio floor — see InferenceConfig::reasoning_floor.

defaults_origin: Option<DefaultsOrigin>

Whether the model’s inference_defaults were user-set or auto-detected. None when the model has no stored inference_defaults at all, in which case it has no effect either way. See DefaultsOrigin.

Implementations

impl ModelSamplingContext

pub fn for_model(model: &Model) -> Self

Read both facts off a catalog row.

Four call sites across three crates built this struct field-by-field from a Model, each re-deriving is_reasoning from the tag list inline. Two fields is exactly the size at which hand-construction looks harmless and stops being so: the pair travels together, both are read by the same fold, and a call site that filled one and defaulted the other would resolve against the wrong floor or mis-rank the model’s own defaults against global settings — silently, in both cases.

Trait Implementations

impl Clone for ModelSamplingContext

fn clone(&self) -> ModelSamplingContext

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ModelSamplingContext

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

Formats the value using the given formatter.

impl Default for ModelSamplingContext

fn default() -> ModelSamplingContext

Returns the “default value” for a type.

impl PartialEq for ModelSamplingContext

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

Tests for self and other values to be equal, and is used by ==.

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

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl Copy for ModelSamplingContext

impl Eq for ModelSamplingContext

impl StructuralPartialEq for ModelSamplingContext