# Deep Reactive Policy (IMPACT) pretraining data

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

- id: `deep-reactive-policy-impact-pretraining-data`
- kind: synthetic or generated
- robot or device: manipulator arm in simulation (class: single arm)
- how collected: generated expert trajectories
- size as stated: 10 million generated expert trajectories
- hours: not stated | hours only claimed: none | episodes: 10,000,000
- year: 2025
- organisation: Carnegie Mellon University
- licence: not stated (class: not stated)
- access: not stated (class: not stated)
- link (paper): https://arxiv.org/abs/2509.06953
- note: Only the paper was opened.

Confirm the licence at the link before relying on it.