agents.vectordbs

agents.vectordbs#

The following vector DB specification classes are meant to define a comman interface for initialization of vector DBs. Currently the only supported vector DB is Chroma.

Module Contents#

Classes#

ChromaDB

Chroma is the open-source AI application database. It provides embeddings, vector search, document storage, full-text search, metadata filtering, and multi-modal retreival support.

API#

class agents.vectordbs.ChromaDB#

Bases: agents.vectordbs.DB

Chroma is the open-source AI application database. It provides embeddings, vector search, document storage, full-text search, metadata filtering, and multi-modal retreival support.

Parameters:

  • username (Optional[str], optional) – The username for authentication. Defaults to None.

  • password (Optional[str], optional) – The password for authentication. Defaults to None.

  • embeddings (str, optional) – Embedding backend to use. Choose from “ollama” or “sentence-transformers”.

  • checkpoint (str, optional) – The model checkpoint to use for embeddings. For example, “bge-large:latest”.

  • ollama_host (str, optional) – Host address for the Ollama service (used if embeddings=“ollama”).

  • ollama_port (int, optional) – Port number for the Ollama service.

  • init_timeout (int, optional) – The timeout in seconds for the initialization process. Defaults to 10 minutes (600 seconds).

To use ChromaDB with the supported embedding backends, install the following Python packages:

pip install ollama  # if using ollama (requires separate Ollama runtime)
pip install sentence-transformers  # if using sentence-transformers

If using the Ollama backend, make sure the Ollama server is running and accessible at the specified host and port.

Example Usage:

from agents.vectordbs import ChromaDB

chroma_config = ChromaDB(
    embeddings='ollama',
    checkpoint='bge-large:latest',
    ollama_host='localhost',
    ollama_port=11434
)
get_init_params() → Dict#

Get init params from models

Contents