This reference describes unreleased code on main; no release documents it yet.

rlmesh::adapters::Concat

structPreviewuse rlmesh::adapters::Concat;[source]

A multi-part numeric state input: several roles packed into one tensor.

struct Concat

parts

field[source]
parts: Vec<ConcatPart>

Roles (or State / Constant parts) concatenated in order. At least one part must carry a role – a state of only constants reads nothing from the env.

pad_to

field[source]
pad_to: Option<u32>

Zero-pad the concatenated vector to this length. Padding is the last step: every part is converted, ranged and scaled, the parts are concatenated in order, and only then is the result padded.

dtype

field[source]
dtype: Tolerant<DType>

Element dtype name of the resulting value.

reshape

field[source]
reshape: Option<Vec<i64>>

Optional target shape for the resulting value.

container

field[source]
container: StateContainer

Emit a tensor or a list of values.

clip

field[source]
clip: Option<(f64, f64)>

Clamp the concatenated vector to (low, high) after every part’s own transforms and before pad_to.

new

associated function[source]
fn new<T: Into<ConcatPart>>(parts: impl IntoIterator<Item = T>) -> Self

Construct a concatenation producing float32 tensors without padding or clipping.

Trait implementations

  • impl Clone for Concat
  • impl Debug for Concat
  • impl StructuralPartialEq for Concat
  • impl PartialEq for Concat
  • impl Serialize for Concat
  • impl<'de> Deserialize<'de> for Concat
  • impl Display for Concat
  • impl From<Concat> for InputNode