ann_benchmark.yaml

float:
  - name: nmslib
    library: nmslib
    method: [hnsw]
    space: [cosinesimil,l2,l1]
    run_groups:
      K:
        query_args: [[10,50,100,300,400,500,1000,2000,2500,3000]]

  - name: hnswlib
    library: hnswlib
    method: [hnswlib]
    space: [cosine,l2,ip]
    run_groups:
      K:
        query_args: [[10,50,100,300,400,500,1000,2000,2500,3000]]
      ef_construction:
        query_args: [[10, 20, 40, 80, 120, 200, 400, 600, 800]]

  - name: bruteforce blas
    library: sklearn.neighbors
    algorithm: [brute]
    metric: [minkowski,euclidean,cosine,manhattan,cityblock]
    run_groups:
      K:
        query_args: [[10,50,100,300,400,500,1000,2000,2500,3000]]

  - name: balltree
    library: sklearn.neighbors
    metric: [minkowski,euclidean,l2,manhattan,cityblock]
    run_groups:
      K:
        query_args: [[10,50,100,300,400,500,1000,2000,2500,3000]]
      leaf_size:
        query_args: [[10, 20, 40, 100, 200, 400, 1000]]

  - name: kdtree
    library: sklearn.neighbors
    metric: [minkowski,euclidean,l2,manhattan,cityblock]
    run_groups:
      K:
        query_args: [[10,50,100,300,400,500,1000,2000,2500,3000]]
      leaf_size:
        query_args: [[10, 20, 40, 100, 200, 400, 1000]]

  - name: ckdtree
    library: scipy.spatial
    run_groups:
      K:
        query_args: [[10,50,100,300,400,500,1000,2000,2500,3000]]
      leaf_size:
        query_args: [[10, 20, 40, 100, 200, 400, 1000]]

  - name: annoy
    library: annoy
    metric: [euclidean,manhattan]
    run_groups:
      K:
        query_args: [[10,50,100,300,400,500,1000,2000,2500,3000]]
      n_trees:
        query_args: [[10,100, 200, 400, 1000, 2000, 4000, 10000, 20000, 40000, 100000,200000, 400000]]

  - name: faiss
    library: annoy
    metric: [l2]
    run_groups:
      K:
        query_args: [[10,50,100,300,400,500,1000,2000,2500,3000]]

  - name: rpforest
    library: rpforest
    run_groups:
      K:
        query_args: [[10,50,100,300,400,500,1000,2000,2500,3000]]

  - name: nearpy
    library: nearpy
    metric: [euclidean]
    run_groups:
      K:
        query_args: [[10,50,100,300,400,500,1000,2000,2500,3000]]