Struct DeclaredPrecisionPlan 

pub struct DeclaredPrecisionPlan {
    pub source: PlanSource,
    pub rules: Vec<Rule>,
    pub ignore: Vec<Target>,
}

2026-09-28: The per-layer precision a checkpoint declares.

Fields

source: PlanSource

2026-09-28: Dialect.

rules: Vec<Rule>

2026-09-28: Schemes.

ignore: Vec<Target>

2026-09-28: Modules left unquantized.

Implementations

impl DeclaredPrecisionPlan

pub const UNDECLARED: DeclaredPrecisionPlan

2026-09-28: A plan that declares nothing (no quantization_config).

pub fn from_quantization_config(qc: &Value) -> Result<Self>

2026-09-28: Parse the quantization_config object (the value, not the whole config.json). A ModelOpt hf_quant_config.json nested under "quantization" is accepted as is.

pub fn resolve(&self, module: &str) -> LayerPrecision

2026-09-28: What module (a checkpoint module path such as model.language_model.layers.3.mlp.gate_proj, no .weight suffix) declares.

Ignored modules are unquantized. Otherwise the compressed-tensors precedence applies: exact names first, then patterns ordered by their text, then class targets (Linear), and the first match wins.

pub fn is_undeclared(&self) -> bool

2026-09-28: True when no module declares anything.

Trait Implementations

impl Clone for DeclaredPrecisionPlan

fn clone(&self) -> DeclaredPrecisionPlan

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for DeclaredPrecisionPlan

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

Formats the value using the given formatter.

impl Default for DeclaredPrecisionPlan

fn default() -> DeclaredPrecisionPlan

Returns the “default value” for a type.