Embedding Search

The /v1/search/embedding endpoint (API) takes an embedding, typically an image or text query that you passed through your own LLM embedding process, and semantic results are returned with their IDs and relevance score. This endpoint is used when the collection being searched has User Provided Embeddings (UPE) or you run some embedding pre-processing on the search query before sending it to the Vantage platform.

👍

Standard Search Options apply

Filtering, pagination, accuracy etc all apply to this endpoint

This example JSON uses this endpoint to tell the Vantage platform what to search and return. The embedding field is used to search for similarity. It is an array with a length of Dimension Size as specified during the collection creation step.

{
  "embedding": [0.110183, -0.3817298, ... ], 
  "request_id": 333666, 
  "collection": {
    "account_id": "docs-account",
    "collection_id": "docs-sample-collection",
    "accuracy": "0.2"
  },
  "filter": {
    "boolean_filter": ""
  },
  "pagination": {
    "page": 0,
    "count": 40
  }
}

📘

Reference Guide for Embedding Search