Prev Next

Database / ChromaDB Interview Questions

What distance metrics does ChromaDB support and how do you choose between them?

ChromaDB uses a distance metric to measure how similar two vectors are during nearest-neighbour search. The metric is set at collection creation time and cannot be changed afterward. Choosing the wrong metric for your embedding model can significantly degrade search quality.

ChromaDB distance metrics
Metrichnsw:space valueFormulaBest for
L2 (Euclidean)l2 (default)√Σ(aáµ¢−báµ¢)²When vector magnitude matters; general purpose
Cosine similaritycosine1 − (a·b)/(-)Text embeddings — focuses on direction not magnitude
Inner Productip−(a·b)When embeddings are pre-normalised to unit length
import chromadb

# Set metric at collection creation — cannot change later!
collection_cosine = client.create_collection(
    name="text_cosine",
    metadata={"hnsw:space": "cosine"},  # recommended for text
)

collection_l2 = client.create_collection(
    name="general_l2",
    metadata={"hnsw:space": "l2"},  # default if not specified
)

collection_ip = client.create_collection(
    name="normalised_ip",
    metadata={"hnsw:space": "ip"},  # use when vectors are unit-normalised
)

# Query returns "distances" field — interpretation depends on metric:
# cosine: 0 = identical, 2 = opposite (lower = more similar)
# l2:     0 = identical, larger = more different (lower = more similar)
# ip:     more negative = more similar (with normalised vectors)

Rule of thumb: most popular text embedding models (OpenAI, Sentence Transformers) are optimised for cosine similarity. Use "hnsw:space": "cosine" for text RAG applications. L2 is the default but is less optimal for text embeddings that vary in magnitude.

Which distance metric is generally recommended for text embedding models like those from OpenAI or Sentence Transformers?
When can you change the distance metric of an existing ChromaDB collection?

More Related questions...

What is ChromaDB and what problem does it solve? What are embeddings and why are they central to how ChromaDB works? What distance metrics does ChromaDB support and how do you choose between them? What is a ChromaDB collection and how do you create, list, get, and delete collections? How do you add documents to a ChromaDB collection? How do you query a ChromaDB collection for similar documents? How do you retrieve, update, and delete specific documents in ChromaDB? How do you filter query results using metadata in ChromaDB? What is the difference between ChromaDB's in-memory and persistent storage modes? What is ChromaDB's default embedding function and how does it work? How do you use the OpenAI embedding function with ChromaDB? How do you use HuggingFace models as embedding functions in ChromaDB? How do you create a custom embedding function for ChromaDB? How does ChromaDB's PersistentClient store data on disk, and what are its limitations? What is the HNSW index in ChromaDB and what parameters can you tune? How do you efficiently add large numbers of documents to ChromaDB using batching? What is the where_document filter in ChromaDB and how does it differ from where? How do you control what data ChromaDB returns in query and get results using include? How do you design metadata schemas for effective filtering in ChromaDB? How do you inspect a ChromaDB collection's contents and configuration? How do you build a basic RAG (Retrieval-Augmented Generation) pipeline with ChromaDB? What are effective document chunking strategies when indexing documents into ChromaDB for RAG? How do you use ChromaDB as a vector store with LangChain? How do you implement multi-tenancy or data isolation in ChromaDB? What is embedding consistency and why is it critical in ChromaDB applications? How do you run ChromaDB as a standalone HTTP server and connect to it from multiple clients? When should you use upsert() instead of add() in ChromaDB, and what are common patterns? What are best practices for structuring ChromaDB collection metadata for production use? How does ChromaDB compare to FAISS, and when should you choose one over the other? What are common ChromaDB errors and how do you handle them in production code? How do you back up and restore a ChromaDB persistent database? How do you ensure the correct embedding function is used when reopening a persistent ChromaDB collection? How do you interpret ChromaDB query distances and convert them into meaningful relevance scores? What are ChromaDB's practical size limits and performance characteristics at scale? How do you use ChromaDB to detect and remove near-duplicate or semantically similar documents? How do you reset or clear a ChromaDB collection without deleting and recreating it? What configuration settings does ChromaDB support and how do you disable telemetry? What is a production readiness checklist for a ChromaDB-based application?
Show more question and Answers...

pgvector basics Interview Questions

Comments & Discussions

Interviews Questions

About Javapedia.net

Javapedia.net is for Java and J2EE developers, technologist and college students who prepare of interview. Also this site includes many practical examples.

This site is developed using J2EE technologies by Steve Antony, a senior Developer/lead at one of the logistics based company.

Follow Javapedia updates

contact: javatutorials2016[at]gmail[dot]com

Kindly consider donating for maintaining this website. Thanks.

Copyright © 2026, javapedia.net, all rights reserved. · .

DMCA.com Protection Status