Go package: cagra
Sources: go/cagra
Source: go/cagra/index_params.go:23
Source: go/cagra/search_params.go:28
Source: go/cagra/search_params.go:19
Source: go/cagra/index_params.go:21
Cagra ANN Index
Source: go/cagra/cagra.go:14
Supplemental parameters to build CAGRA Index
Source: go/cagra/index_params.go:17
Parameters to extend CAGRA Index
Source: go/cagra/extend_params.go:11
Source: go/cagra/search_params.go:26
Source: go/cagra/index_params.go:12
Source: go/cagra/search_params.go:17
Supplemental parameters to search CAGRA Index
Source: go/cagra/search_params.go:13
Builds a new Index from the dataset for efficient search.
Source: go/cagra/cagra.go:38
Creates a new CompressionParams
Source: go/cagra/index_params.go:36
Creates a new ExtendParams
Source: go/cagra/extend_params.go:16
Creates a new empty Cagra Index
Source: go/cagra/cagra.go:20
Creates a new IndexParams
Source: go/cagra/index_params.go:99
Creates a new SearchParams
Source: go/cagra/search_params.go:35
Extends the index with additional data
Source: go/cagra/cagra.go:55
Perform a Approximate Nearest Neighbors search on the Index
Source: go/cagra/cagra.go:85
Destroys the Cagra Index
Source: go/cagra/cagra.go:67
The number of iterations searching for kmeans centers (both VQ & PQ phases).
Source: go/cagra/index_params.go:76
The bit length of the vector element after compression by PQ.
Source: go/cagra/index_params.go:52
The dimensionality of the vector after compression by PQ. When zero, an optimal value is selected using a heuristic.
Source: go/cagra/index_params.go:60
The fraction of data to use during iterative kmeans building (PQ phase). When zero, an optimal value is selected using a heuristic.
Source: go/cagra/index_params.go:92
The fraction of data to use during iterative kmeans building (VQ phase). When zero, an optimal value is selected using a heuristic.
Source: go/cagra/index_params.go:84
Vector Quantization (VQ) codebook size - number of “coarse cluster centers”. When zero, an optimal value is selected using a heuristic.
Source: go/cagra/index_params.go:68
Source: go/cagra/extend_params.go:40
The additional dataset is divided into chunks and added to the graph. This is the knob to adjust the tradeoff between the recall and operation throughput. Large chunk sizes can result in high throughput, but use more working memory (O(max_chunk_size*degree^2)). This can also degrade recall because no edges are added between the nodes in the same chunk. Auto select when 0.
Source: go/cagra/extend_params.go:35
Destroys IndexParams
Source: go/cagra/index_params.go:152
ANN algorithm to build knn graph
Source: go/cagra/index_params.go:126
Compression parameters
Source: go/cagra/index_params.go:145
Degree of output graph
Source: go/cagra/index_params.go:119
Degree of input graph for pruning
Source: go/cagra/index_params.go:113
Number of iterations to run if building with NN_DESCENT
Source: go/cagra/index_params.go:138
Destroys SearchParams
Source: go/cagra/search_params.go:157
Which search implementation to use.
Source: go/cagra/search_params.go:67
Upper limit of hashmap fill rate. More than 0.1, less than 0.9.
Source: go/cagra/search_params.go:139
Lower limit of hashmap bit length. More than 8.
Source: go/cagra/search_params.go:133
Hashmap type. Auto selection when AUTO.
Source: go/cagra/search_params.go:113
Number of intermediate search results retained during the search. This is the main knob to adjust trade off between accuracy and search speed. Higher values improve the search accuracy
Source: go/cagra/search_params.go:55
Upper limit of search iterations. Auto select when 0.
Source: go/cagra/search_params.go:61
Maximum number of queries to search at the same time (batch size). Auto select when 0
Source: go/cagra/search_params.go:47
Lower limit of search iterations.
Source: go/cagra/search_params.go:95
Number of iterations of initial random seed node selection. 1 or more.
Source: go/cagra/search_params.go:145
Bit mask used for initial random seed node selection.
Source: go/cagra/search_params.go:151
How many nodes to search at once. Auto select when 0.
Source: go/cagra/search_params.go:101
Number of threads used to calculate a single distance. 4, 8, 16, or 32.
Source: go/cagra/search_params.go:89
Thread block size. 0, 64, 128, 256, 512, 1024. Auto selection when 0.
Source: go/cagra/search_params.go:107