Hello, When I do a K=1 dense vector query and copy the vector of an existing document as the target vector (let's say [1.0, 2.0, 3.0]), I don't get the same document back in the result (which I'd expect as its distance to the query is 0). I understand that the HNSW graph can result in a local minima, but this still seems strange to me that such a "good match" would be absent from the result. Has anyone else experienced this phenomena?
It is also worth noting that when we query K=10, we do get the exact match document as the top result and the previously returned single document is nowhere in the result set. -Kevin Liang
