Encyclopedia Evalica / Evaluation / Reference-based scoring

Reference-based scoring

/'reh.fer.uhns bayst 'skaw.rihng/Scoring that compares output to a ground truth/expected output. It works best when "correctness" is well-defined. (noun)

We use reference-based scoring for extraction tasks with known answers.

Related Evaluation terms

From the docs

Get started with Evals

Braintrust is the observability platform for agents in production. By actively applying intelligence to agent traces and automatically surfacing critical patterns, Braintrust helps teams at Notion, Stripe, Box, OpenAI, and Cloudflare ship quality agents at scale.

Start building