kasinathansj/text-embedding

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

Embedding API

This is a FastAPI-based embedding server that uses the multi-qa-mpnet-base-dot-v1 model from SentenceTransformers to generate text embeddings.

Model Specification

• —Model Name: multi-qa-mpnet-base-dot-v1
• —Architecture: MPNet (Microsoft Pretrained Network)
• —Embedding Size: 768
• —Use Case: Optimized for multi-question answering (QA) and semantic search
• —Number of Layers: 12
• —Number of Attention Heads: 12
• —Hidden Size: 768
• —Intermediate Size: 3072
• —Vocabulary Size: 30,527
• —Framework: SentenceTransformers (Hugging Face)

API Endpoints

1. Health Check

Endpoint: GET /

Description: Checks if the embedding server is running.

Response:

{
  "message": "Embedding server is running!"
}

2. Generate Embedding

Endpoint: POST /embed/

Description: Generates an embedding for the provided input text.

Request Body:

{
  "text": "Sample input text"
}

Response:

{
  "embedding": [0.12345, -0.6789, ...] // 768-dimensional vector
}

Running the API

To start the API server, run:

uvicorn main:app --host 0.0.0.0 --port 8000

Installation

1. 1.Install dependencies:

pip install fastapi uvicorn sentence-transformers

1. 1.Make sure the model is available in ./saved_model/.
2. 2.Start the server as mentioned above.

Usage

You can test the API using curl, Postman, or any HTTP client:

curl -X POST "http://localhost:8000/embed/" \
     -H "Content-Type: application/json" \
     -d '{"text": "Hello world"}'

This will return the 768-dimensional embedding for the text input.

Notes

• —Ensure that ./saved_model contains the pre-trained multi-qa-mpnet-base-dot-v1 model.
• —The API is optimized for embedding short text snippets for semantic search and retrieval tasks.