Struct Model 

pub struct Model { /* private fields */ }

An ML inference model session.

Wraps a TFLM MicroInterpreter with automatic cleanup.

Implementations

impl Model

pub fn new(config: &ModelConfig<'_>) -> Result<Self>

Create a new model via heap allocation.

pub unsafe fn from_static( storage: *mut ove_model_storage_t, arena: *mut u8, config: &ModelConfig<'_>, ) -> Result<Self>

Create from caller-provided storage and arena.

Available in both heap and zero-heap modes. Useful when the same storage/arena must be reused for different models (e.g. two-stage inference pipelines).

Safety

Caller must ensure storage and arena outlive the Model and are not shared with another primitive.

pub fn invoke(&self) -> Result<()>

Run the model forward pass.

pub fn input(&self, index: u32) -> Result<TensorInfo>

Get tensor info for an input tensor.

pub fn output(&self, index: u32) -> Result<TensorInfo>

Get tensor info for an output tensor.

pub fn last_inference_us(&self) -> u64

Return last inference duration in microseconds.

impl Model

pub fn input_slice_mut<T>(&self, index: u32) -> Result<&mut [T]>

Get input tensor data as a mutable typed slice.

The slice length is tensor_info.size / size_of::<T>().

Errors

Returns an error if the tensor index is invalid.

pub fn output_slice<T>(&self, index: u32) -> Result<&[T]>

Get output tensor data as a typed slice.

The slice length is tensor_info.size / size_of::<T>().

Errors

Returns an error if the tensor index is invalid.

Trait Implementations

impl Debug for Model

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

Formats the value using the given formatter.

impl Drop for Model

fn drop(&mut self)

Executes the destructor for this type.

impl Send for Model

impl Sync for Model