The Chroma Vector Store node allows you to store and query vectorized documents in a ChromaDB server. It connects to your specified host and port, using a defined collection name to organize indexed data.
Inputs
- Documents – Accepts vectorized document chunks for storage. These typically come from embedding nodes.
- Questions – Accepts queries (e.g., user prompts or questions) to search the vector database for relevant context.
Outputs
- Answers – Returns context-relevant document results retrieved from the vector store in response to input questions.
- Documents – Emits the full document metadata or ID associated with retrieved vector matches.
- Questions (output) – Returns the question passed through after retrieval, useful for chaining with LLM nodes.
Configuration Steps
- Type of Chroma Host – Choose the environment in which your ChromaDB instance is hosted.
- Example – Your own ChromaDB server
- Host – Specify the server address to connect to Chroma.
- Example – localhost
- Port – Define the port where Chroma is listening.
- Example – 8330
- Collection – Set the collection name where your documents will be stored and queried.
- Example – APARAVI
