Struct TaskBatch 

pub struct TaskBatch<A: Float + ScalarOperand + Debug> {
    pub initial_params: Array<A, IxDyn>,
    pub inner_gradients: Vec<Array<A, IxDyn>>,
    pub final_loss_grad: Array<A, IxDyn>,
}

Per-task data used by the meta-update.

For each task T_i the caller supplies:

• initial_params – the meta-parameters theta at the start of the inner loop (theta_i^{(0)}).
• inner_gradients – the gradients g_0, g_1, ..., g_{K-1} evaluated at each inner iterate theta_i^{(0)}, ..., theta_i^{(K-1)}. The length of this vector must be at least 1 and matches the number of inner steps.
• final_loss_grad – the gradient of the meta-loss evaluated at the adapted parameters theta_i^{(K)} (sometimes denoted ∂L/∂theta').

All arrays must share the same shape; mismatches surface as OptimError::InvalidParameter.

Fields

initial_params: Array<A, IxDyn>

Parameters at the start of the inner loop (theta_i^{(0)}).

inner_gradients: Vec<Array<A, IxDyn>>

Sequence of gradients along the inner trajectory.

final_loss_grad: Array<A, IxDyn>

Gradient of the meta-loss at the adapted parameters.

Trait Implementations

impl<A: Clone + Float + ScalarOperand + Debug> Clone for TaskBatch<A>

fn clone(&self) -> TaskBatch<A>

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<A: Debug + Float + ScalarOperand + Debug> Debug for TaskBatch<A>

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

Formats the value using the given formatter.