Dyna-2 trained on 1M+ hours of egocentric video; data infra detailed, open-sourced
Scobleizer · x · 2026-08-19
Robotics firm Dyna published how they trained Dyna-2 repeatably on over 1,000,000 hours of egocentric video, noting that what worked at 10k hours broke at scale: ingestion capped at 14,000 episode-hours/week (a million hours would take over a year), training manifests took 48 hours to build, and reading a petabyte from cloud storage exposed GPUs to latency and packet loss.
A YC team resharing the thread called it a gold mine and open-sourced HFlow, a reference implementation of the described infrastructure: Airflow for DAG orchestration, MCAP for episode storage with random access, and DuckDB for analytical queries and dataset curation over columnar metadata. Their takeaway after two months at YC: at scale, AI problems become software engineering problems—fix one bottleneck and it moves elsewhere.
More from Embodied
- CMU's Andrea Bajcsy on what safety means for generalist robots, beyond collision avoidance — CSProfKGD · 2026-10-16
- Codex drives a physical CAN-bus motor, eyeing a fully Codex-built robot — OpenAIDevs · 2026-10-16
- Berkeley paper uses the Midas myth to rethink robotics beyond human imitation — begusgasper · 2026-10-16
- Team uses Claude Opus 5.5 to reproduce core claims of 10 recent robotics papers end to end — teortaxesTex · 2026-10-16
- Kinex: a harness that connects coding agents to any physical robot for scalable learning — daibond_alpha · 2026-10-16
- Gemini is a robotics outlier: dropping it lifts capability-robotics correlation from 0.20 to 0.96 — ycombinator · 2026-10-16