Embeddings

Embeddings measure the relatedness of text strings. They can be used in:

• Search: to rank results by relevance to query string
• Clustering: to group strings by similarity
• Recommendations: to recommend items with related text strings
• Anomaly detection: to identify outliers
• Diversity measurement: to analyze similarity distribution
• Classification: to classify text strings by their labels

Embeddings are floating-point number vectors where the distance between them indicates relatedness. Shorter distances signify greater relatedness, while longer distances suggest less.

Getting Embeddings

To get an embedding, send a text string and a model name to the embedding API endpoint, for example:

curl -X 'POST' \
  'https://deepfellow-server-host/v1/embeddings' \
  -H "Authorization: Bearer DEEPFELLOW-PROJECT-API-KEY" \
  -H 'Content-Type: application/json' \
  -d '{
  "input": "Your text",
  "model": "mxbai-embed-large"
}'

from openai import OpenAI
client = OpenAI(
    base_url="https://deepfellow-server-host/v1",
    api_key="DEEPFELLOW-PROJECT-API-KEY"
)

response = client.embeddings.create(
    input="Your text",
    model="mxbai-embed-large"
)

print(response.model_dump_json())

import OpenAI from 'openai';

const client = new OpenAI({
    baseURL: 'https://deepfellow-server-host/v1',
    apiKey: 'DEEPFELLOW-PROJECT-API-KEY'
});

const response = await client.embeddings.create({
    input: 'Your text',
    model: 'mxbai-embed-large'
});

console.log(response);

The input parameter accepts a single string or an array of strings, letting you embed multiple inputs in a single request.

The response includes an embedding vector, which is a list of floating-point numbers, as well as some extra metadata. This vector can be extracted and stored in a vector database for further use.

{
    "object": "list",
    "data": [
        {
            "object": "embedding",
            "index": 0,
            "embedding": [
                -0.006929283495992422, -0.005336422007530928, ..., -4.547132266452536555,
            ]
        }
    ],
    "model": "mxbai-embed-large",
    "usage": {
        "prompt_tokens": 2,
        "total_tokens": 2
    }
}

Optional Parameters

You can configure the following optional request parameters:

• dimensions: The number of dimensions the resulting output embeddings will have, while preserving their ability to represent concepts.
• encoding_format: The format to return the embeddings in, for example float.
• user: A unique identifier for the end-user.

See API Reference for details.

To learn more using embeddings in different use-cases, visit OpenAI API Documentation.