embedding

A small Rust HTTP service for generating vector embeddings.

Quick start

docker compose up --build

First request triggers the model download into ./models (bind-mounted into the container); subsequent restarts reuse it.

curl -X POST http://localhost:3000/embed \
  -H 'content-type: application/json' \
  -d '{"texts":["hello world","another piece of text"]}'

Configuration

Configured via environment variables (set them in .env):

Variable	Default	Description
EMBEDDING_PORT	3000	Port the service listens on.
EMBEDDING_MODELS	nomic	Comma-separated list of models to load.
EMBEDDING_CACHE_DIR	(default cache)	Directory for downloaded model files.
EMBEDDING_POOL_SIZE	(memory-derived)	Number of model instances per pool.

API

POST /embed

Request:

{ "texts": ["string", "..."] }

Response:

{
  "model": "NomicEmbedTextV15",
  "embeddings": [[0.012, -0.034, ...], ...],
  "tokens": 17
}

Errors:

• 400 Bad Request — texts is empty.
• 500 Internal Server Error — embedding or tokenizer failure (message in error field).