// Docs
Refiner Documentation
Refiner is our open-source library for reading, transforming, and writing robotics datasets.
These docs cover Refiner’s data-processing utilities and reference implementations from our published research.
For access to our proprietary models and pipelines, contact us directly. Send us a small video sample and we’ll show you what we can extract.
// 01 — COMMUNITY
// 02 — GET STARTED
// 03 — ALL TOPICS
- Running pipelines
How Refiner pipelines execute on your own infrastructure
- Reading data
Choose and configure Refiner readers for robotics data
- Episode data
How Refiner represents robotics episodes, frames, videos, metadata, tasks, and stats
- Transforms
How to change, enrich, filter, and type episode rows in Refiner
- Episode operations
Higher-level operations for robotics episode data
- Inference
Use language, vision-language, structured, vLLM, and pooling inference in Refiner
- Writing data
Write robotics datasets and pipeline outputs from Refiner
- Examples
End-to-end Refiner workflows for robotics data
- Reference
Reference material for Refiner paths, dependencies, and public APIs