Trait BackendExecutionContext

pub trait BackendExecutionContext: Send + Sync {
    // Required methods
    fn set_input(&mut self, id: Id, tensor: &Tensor) -> Result<(), BackendError>;
    fn get_output(&mut self, id: Id) -> Result<Tensor, BackendError>;
    fn compute(
        &mut self,
        inputs: Option<Vec<NamedTensor>>,
    ) -> Result<Option<Vec<NamedTensor>>, BackendError>;
}

A BackendExecutionContext performs the actual inference; this is the backing implementation for a user-facing execution context.

Required Methods

fn set_input(&mut self, id: Id, tensor: &Tensor) -> Result<(), BackendError>

fn get_output(&mut self, id: Id) -> Result<Tensor, BackendError>

fn compute( &mut self, inputs: Option<Vec<NamedTensor>>, ) -> Result<Option<Vec<NamedTensor>>, BackendError>

Implementors