pub type SparseVector = Vec<(u32, f32)>;Expand description
A single learned-sparse vector as (term_id, weight) pairs.
Each term_id indexes into the model’s term space (see
SparseEmbeddingModel::vocab_size); weight is the non-negative learned
importance of that term. Entries are sparse: only non-zero terms appear.
Term ids are only comparable within the same model — two models with
different vocabularies produce incompatible sparse vectors.
Aliased Type§
pub struct SparseVector { /* private fields */ }