Struct StreamingConversionPlan 

pub struct StreamingConversionPlan { /* private fields */ }

Metadata, pinned source digest, and per-tensor policy for one bounded conversion.

A real model recipe supplies one TensorConversion for every tensor named by its WeightsManifest. The plan contains no source payload and no machine-specific packing; it can therefore be reviewed before a multi-gigabyte conversion opens an output file.

Implementations

impl StreamingConversionPlan

pub fn new( model_family: impl Into<String>, source_sha256: impl Into<String>, ) -> Self

Starts a portable conversion plan tied to the expected source SHA-256.

pub fn license_notice(self, notice: impl Into<String>) -> Self

Sets the required Apache-2.0 attribution and change notice.

pub fn model_config(self, config: Value) -> Self

Records the frozen source model configuration in the artifact directory.

pub fn quantization_manifest(self, manifest: Value) -> Self

Records the reviewed per-tensor quantization recipe in the artifact directory.

pub fn tensor(self, tensor: TensorConversion) -> Self

Adds one explicit source-to-artifact tensor conversion.

Trait Implementations

impl Clone for StreamingConversionPlan

fn clone(&self) -> StreamingConversionPlan

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for StreamingConversionPlan

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

Formats the value using the given formatter.