index_put_(input: Tensor, indices: list[Tensor] | tuple[Tensor, ...], values: Tensor, accumulate: bool = False)In-place variant of index_put — mutates input so the
write is visible through the same Tensor reference.
Internally this rebinds input._impl to the freshly-built result,
matching the convention Lucid already uses for engine-level in-place
ops (add_, mul_). Storage-level mutation is not currently
available for composite indexing — autograd consumers should treat
the returned tensor as a new node.
Parameters
inputTensorDestination; mutated in place via
_impl rebind.Per-axis index tensors (one per dimension of
input); same
contract as index_put.valuesTensorValues to write at the addressed positions.
accumulatebool= FalseWhen
True add to existing values (duplicate indices sum);
when False (default) overwrite (duplicate indices resolve
to the last write).Returns
TensorThe same input tensor, now holding the updated values.