# DreamDojo-HV

From the catalogue. Listed by one researcher from one source page.

- id: `dreamdojo-hv`
- kind: egocentric, video for world models
- robot or device: egocentric human videos (class: human egocentric)
- how collected: aggregated for world model pretraining
- size as stated: 44k hours
- hours: 44,000 | hours only claimed: none | episodes: not stated
- year: 2026
- organisation: Shenyuan Gao et al.
- licence: Apache-2.0 (code) (class: unclear)
- access: open (class: open)
- link (paper): https://arxiv.org/abs/2602.06949
- other names: DreamDojo data (DreamDojo-HV and GR-1 post-training sets)
- note: Pretraining mixture.

Confirm the licence at the link before relying on it.