Struct RagConfig 

pub struct RagConfig {
    pub vector_weight: f64,
    pub graph_weight: f64,
    pub top_k_candidates: usize,
    pub top_k_rerank: usize,
    pub enable_graph_expansion: bool,
    pub graph_depth: u32,
    pub context_depth: u32,
    pub max_context_entities: usize,
    pub min_vector_score: f32,
    pub min_combined_score: f64,
    pub vector_dimension: usize,
}

Configuration knobs for the RAG pipeline. All fields have defaults tuned to the paper recommendations.

Fields

vector_weight: f64

Weight applied to the vector (semantic) score. Default: 0.6.

graph_weight: f64

Weight applied to the graph connectivity score. Default: 0.4.

top_k_candidates: usize

Stage-1: how many ANN candidates to fetch from TurboQuant. Default: 50.

top_k_rerank: usize

Stage-2: how many candidates survive after exact rerank. Default: 20.

enable_graph_expansion: bool

Whether to expand candidates via BFS neighbours. Default: true.

graph_depth: u32

BFS depth for graph-score neighbour collection. Default: 1.

context_depth: u32

BFS depth for context collection. Default: 2.

max_context_entities: usize

Maximum context entities attached to each result. Default: 5.

min_vector_score: f32

Minimum vector score for a candidate to survive. Default: 0.0 (off).

min_combined_score: f64

Minimum combined score for a result to survive. Default: 0.0 (off).

vector_dimension: usize

Vector dimension; set to match your embedding model. Default: 384.

Trait Implementations

impl Clone for RagConfig

fn clone(&self) -> RagConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for RagConfig

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

Formats the value using the given formatter.

impl Default for RagConfig

fn default() -> Self

Returns the “default value” for a type.