Struct InferencePipeline 

pub struct InferencePipeline { /* private fields */ }

Builder-style inference pipeline that wraps a SequentialModel with optional pre- and post-processing closures.

Implementations

impl InferencePipeline

pub fn new(model: SequentialModel) -> Self

Create a new pipeline wrapping the given model with no pre/post processing.

pub fn with_preprocess<F>(self, f: F) -> Self

where F: Fn(&Tensor) -> Result<Tensor, ModelError> + 'static,

Attach a preprocessing closure that transforms the input tensor before it is fed to the model.

pub fn with_postprocess<F>(self, f: F) -> Self

where F: Fn(&Tensor) -> Result<Tensor, ModelError> + 'static,

Attach a postprocessing closure that transforms the model output tensor before it is returned to the caller.

pub fn run(&self, input: &Tensor) -> Result<Tensor, ModelError>

Run the full pipeline on a single input tensor: preprocess (if set) -> model forward inference -> postprocess (if set).

pub fn run_batch(&self, inputs: &[Tensor]) -> Result<Vec<Tensor>, ModelError>

Run the pipeline on a batch of input tensors, collecting results.