Open
Description
Hi, would it be possible to somehow get the list of data points from the first layer of the HNSW graph?
Context:
- I'm storing HNSW graphs on disk in different files
- When a query comes in, I would like to quickly scan the closest points of the graphs without resorting to the full graph file yet.
- Once I have a short list of the candidate points, I will initiate the HNSW based search of top K in each corresponding file
In my algorithm there is a so-called seed point that is used to group the points into future HNSW files, but probably it is not the best one for the efficient search?
Metadata
Metadata
Assignees
Labels
No labels