Viewing Gentrace v2 documentation
View Gentrace v1
Gentrace home page
Search...
⌘K
Getting started
Get started
Quickstart
Installation
Initialization
Error analysis
Overview
Gentrace Chat
Derivations
Views
Alerting
Tracing
Overview
interaction()
traced()
Instrumentations
Manual LLM instrumentation
Evaluation
Overview
Experiments
Datasets
Unit tests
Dataset tests
Integrations
AI SDK
LangChain
LangGraph
Mastra
Next.js
OpenAI Agents
OpenAI (JS)
OpenAI (Python)
Pydantic AI
Administration
Single sign-on (SSO)
SCIM
Roles
Self-hosted deployment
Reference
OpenTelemetry guide
OpenTelemetry API
SDK Errors
SDK Entities
Datasets
Experiments
Pipelines
Test Cases
Support
Dashboard
Viewing Gentrace v2 documentation
View Gentrace v1
Gentrace home page
Search...
⌘K
Ask AI
Support
Dashboard
Dashboard
Search...
Navigation
Evaluation
Overview
Documentation
API Reference
Documentation
API Reference
Evaluation
Overview
Copy page
Overview of Gentrace’s evaluation system.
Copy page
Gentrace’s agent evaluation operates in three steps:
Create a
dataset
with test cases for your AI pipeline
Run an
experiment
using
unit tests
and/or
dataset tests
Analyze results with
Gentrace Chat
and
derivations
to extract insights and monitor performance
Next steps
Set up
experiments
to run systematic evaluations
Create
datasets
to organize your test cases
Use
derivations
to analyze your results
Previous
Experiments
Create and submit experiments to Gentrace with `experiment()`
Next
On this page
Next steps
Assistant
Responses are generated using AI and may contain mistakes.