Default parameters for CAGRA-style kNN search (deprecated)

The CAGRA, IVF and exhaustive GPU searches share one parameter wrapper now, see params_nn_gpu().

Usage

params_sc_cagra(
  k = 15L,
  ann_dist = "euclidean",
  node_degree_final = NULL,
  k_build = NULL,
  refine_sweeps = 0L,
  max_iters = NULL,
  n_trees = NULL,
  delta = 0.001,
  rho = NULL,
  beam_width = NULL,
  max_beam_iters = NULL,
  n_entry_points = NULL
)

Arguments

k

Integer. Number of neighbours. Carried on the returned list so the deprecated generics can still read it.

ann_dist

Character. One of "euclidean" or "cosine".

node_degree_final

Optional integer. Final node degree of the CAGRA navigational graph.

k_build

Optional integer. Node degree during the NNDescent build phase before CAGRA pruning.

refine_sweeps

Integer. Ignored, the knob is gone.

max_iters

Optional integer. Ignored, the knob is gone.

n_trees

Optional integer. Number of trees in the initial forest.

delta

Numeric. Termination criterion for the NNDescent iterations.

rho

Optional numeric. Sampling rate during NNDescent iterations.

beam_width

Optional integer. Beam width during querying.

max_beam_iters

Optional integer. Maximum beam iterations.

n_entry_points

Optional integer. Number of entry points.

Value

A list with the parameters, as returned by params_nn_gpu().