VectorSearch score is a very large number

MongoDB Atlas Vector Search
1

Hi folks I created a new vector search based index as such:

{
  "fields": [
    {
      "numDimensions": 2048,
      "path": "features",
      "similarity": "dotProduct",
      "type": "vector"
    },
    {
      "path": "content_type",
      "type": "filter"
    },
    {
      "path": "publication_id",
      "type": "filter"
    },
    {
      "path": "issue_date",
      "type": "filter"
    }
  ]
}

This index stores an extracted features of a publication image.

When I ran a sample query with an entry from the index, I’m getting results that are not similar (could be my feature extraction code, I’m also looking into that) however I noticed that the vectorSearch is returning very large distance numbers:
score -> 713253.625
All scores are above 700_000 which made me think if something is wrong with my query index definition.

Any suggestions?

Thank you