Struct ConversionOptions 

pub struct ConversionOptions {
    pub quantization: Option<String>,
    pub verify: bool,
    pub compute_stats: bool,
    pub tolerance: f32,
    pub preserve_metadata: bool,
    pub add_provenance: bool,
    pub tokenizer_path: Option<PathBuf>,
}

Options for conversion operations

Fields

quantization: Option<String>

Target quantization (None = preserve original)

verify: bool

Verify after conversion

compute_stats: bool

Compute tensor statistics

tolerance: f32

Tolerance for floating-point comparison

preserve_metadata: bool

Preserve metadata from source

add_provenance: bool

Add conversion provenance to metadata

tokenizer_path: Option<PathBuf>

External tokenizer.json for weights-only models (PMAT-232)

Implementations

impl ConversionOptions

pub fn without_quantization(&self) -> ConversionOptions

Copy of these options with quantization removed.

Used for the intermediate hop of a multi-step conversion: quantizing at the intermediate step and at the final export would quantize the weights twice.

Trait Implementations

impl Clone for ConversionOptions

fn clone(&self) -> ConversionOptions

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ConversionOptions

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

Formats the value using the given formatter.

impl Default for ConversionOptions

fn default() -> ConversionOptions

Returns the “default value” for a type.