Commit d7190be6 authored by Boris Lukačovič's avatar Boris Lukačovič
Browse files

FIX: missing names for datasets in index specific configs

parent 129c61f8

faiss_ivf_flat.config.yaml

+102 −50
Original line number Diff line number Diff line 
# yaml-language-server: $schema=./config.schema.json



indexProperties:

  - name: m-index

    properties:
@@ -26,7 +28,8 @@ indexProperties: 


experiments:

- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -54,7 +57,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -82,7 +86,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -110,7 +115,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -138,7 +144,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -166,7 +173,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -194,7 +202,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -222,7 +231,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -250,7 +260,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -278,7 +289,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -306,7 +318,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -334,7 +347,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -362,7 +376,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -390,7 +405,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -418,7 +434,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -446,7 +463,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -474,7 +492,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -502,7 +521,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -530,7 +550,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -558,7 +579,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -586,7 +608,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -614,7 +637,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -642,7 +666,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -670,7 +695,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -698,7 +724,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -726,7 +753,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -754,7 +782,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -782,7 +811,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -810,7 +840,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -838,7 +869,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -866,7 +898,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -894,7 +927,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -922,7 +956,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -950,7 +985,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -978,7 +1014,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -1006,7 +1043,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -1034,7 +1072,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -1062,7 +1101,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -1090,7 +1130,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -1118,7 +1159,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -1146,7 +1188,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -1174,7 +1217,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -1202,7 +1246,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -1230,7 +1275,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -1258,7 +1304,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -1286,7 +1333,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -1314,7 +1362,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -1342,7 +1391,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -1370,7 +1420,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5
@@ -1398,7 +1449,8 @@ experiments: 
    name: Delete_balanced



- datasets:

  - datasetType: H5

  - name: Subset of Laion2B

    datasetType: H5

    manipulationDataset:

      key: emb

      path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5

faiss_ivf_flat_26.config.yaml

+8 −4
Original line number Diff line number Diff line
@@ -28,14 +28,16 @@ indexProperties: 


experiments:

  - datasets:

      - datasetType: H5

      - name: Subset of Laion2B

        datasetType: H5

        manipulationDataset:

          key: emb

          path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5

        queryDataset:

          key: emb

          path: datasets/data2024/public-queries-2024-laion2B-en-clip768v2-n=10k.h5

      - datasetType: H5

      - name: Clustered subset of Laion2B

        datasetType: H5

        manipulationDataset:

          key: vectors

          path: datasets/data2024/clustered_laion2B_n=300K_dataset.h5
@@ -196,7 +198,8 @@ experiments: 
            numberOfObjects: 100000



  - datasets:

      - datasetType: H5

      - name: SIFT

        datasetType: H5

        manipulationDataset:

          key: train

          path: datasets/sift/sift-128-euclidean.hdf5
@@ -387,7 +390,8 @@ experiments: 


  # DEBUG - SMALL DATASET FOR DEBUGGING

  - datasets:

      - datasetType: H5

      - name: Subset of Laion2B

        datasetType: H5

        manipulationDataset:

          key: emb

          path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5

faiss_ivf_flat_35.config.yaml

+8 −4
Original line number Diff line number Diff line
@@ -28,14 +28,16 @@ indexProperties: 


experiments:

  - datasets:

      - datasetType: H5

      - name: Subset of Laion2B

        datasetType: H5

        manipulationDataset:

          key: emb

          path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5

        queryDataset:

          key: emb

          path: datasets/data2024/public-queries-2024-laion2B-en-clip768v2-n=10k.h5

      - datasetType: H5

      - name: Clustered subset of Laion2B

        datasetType: H5

        manipulationDataset:

          key: vectors

          path: datasets/data2024/clustered_laion2B_n=300K_dataset.h5
@@ -196,7 +198,8 @@ experiments: 
            numberOfObjects: 100000



  - datasets:

      - datasetType: H5

      - name: SIFT

        datasetType: H5

        manipulationDataset:

          key: train

          path: datasets/sift/sift-128-euclidean.hdf5
@@ -387,7 +390,8 @@ experiments: 


  # DEBUG - SMALL DATASET FOR DEBUGGING

  - datasets:

      - datasetType: H5

      - name: Subset of Laion2B

        datasetType: H5

        manipulationDataset:

          key: emb

          path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5

faiss_ivf_flat_50.config.yaml

+8 −4
Original line number Diff line number Diff line
@@ -28,14 +28,16 @@ indexProperties: 


experiments:

  - datasets:

      - datasetType: H5

      - name: Subset of Laion2B

        datasetType: H5

        manipulationDataset:

          key: emb

          path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5

        queryDataset:

          key: emb

          path: datasets/data2024/public-queries-2024-laion2B-en-clip768v2-n=10k.h5

      - datasetType: H5

      - name: Clustered subset of Laion2B

        datasetType: H5

        manipulationDataset:

          key: vectors

          path: datasets/data2024/clustered_laion2B_n=300K_dataset.h5
@@ -196,7 +198,8 @@ experiments: 
            numberOfObjects: 100000



  - datasets:

      - datasetType: H5

      - name: SIFT

        datasetType: H5

        manipulationDataset:

          key: train

          path: datasets/sift/sift-128-euclidean.hdf5
@@ -387,7 +390,8 @@ experiments: 


  # DEBUG - SMALL DATASET FOR DEBUGGING

  - datasets:

      - datasetType: H5

      - name: Subset of Laion2B

        datasetType: H5

        manipulationDataset:

          key: emb

          path: datasets/data2024/laion2B-en-clip768v2-n=300K.h5

m-tree.config.yaml

+242 −121

File changed.

Preview size limit exceeded, changes collapsed.