Held-out test sets
New data, never published online, for evaluating models without contamination.
Research & Academia
Human data for AI research
Research depends on data you can trust and test sets models have never seen. One6 collects new data, annotates it to written guidelines and staffs research, mathematics, physics and statistics experts for evaluation.
Where human data helps
Expert evaluation
Specialists review model reasoning in mathematics, physics, statistics and other fields.
Annotation for studies
Image, text and audio labeling with written guidelines and agreement measurement.
Speech and language data
Speech and text from native speakers, including lower-resource languages.
Who works on these projects
- Research specialists
- Mathematics, physics and statistics experts
- Native speakers
- Trained annotators working to written guidelines
Relevant services
- AI Training Data CollectionVoice, text, image, video and custom datasets collected from real people to your specification.
- Human Evaluation for AI ModelsDomain experts and trained generalists rate, test and review model output.
- Data Annotation & LabelingImage, video, text and audio labeling with written guidelines and layered quality review.
Have a data project in mind?
Tell us what your model needs. We'll come back with questions, a proposed approach and a pilot plan.