Evaluations Reference

Reference for evaluations objects, views, and common fields

Object Relationships

Key Views

Evaluations Hub

Element	Description
Search bar	Filters datasets by name/query text
Trait filters	Multi-select capability/domain filters
Dataset cards	Show modalities, traits, description, metadata links
Result counter	Shows filtered count vs total

Evaluation Detail

Tab	What It Shows
Details	Dataset description, context, and expectations
Leaderboard	Ranked model results for the dataset
Evaluations Explorer	Sample-level prompt, response, and metric details

Experiments

Field	Meaning
Name	Experiment identifier
Models	Model(s) included in runs
Traits	Traits selected in runs
Status	Current state of experiment activities
Tags	Operator-defined grouping labels
Created date	Experiment creation timestamp

Common Run States

• Running
• Completed
• Failed

Common Metadata and Signals

• Modalities (text, image, video, actions, embeddings)
• Trait associations
• Estimated input/output tokens
• Run timing and duration
• Trait-level benchmark score