'How to search knn in faiss with minimal distance > 0
I am performing a search in faiss, but sometimes the samples i query for are indexed, which is ok. But when querying for k nearest neighbors one of the returned samples is the queried sample. I can remove it post search but then I get k-1 samples which is not what I want. Any way to ignore completely similar samples during the search? by setting the min distance to be > 0 for example? or any other method.
Sources
This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.
Source: Stack Overflow
| Solution | Source |
|---|