🧪 Test?View on arXiv
ZimaBlue: Evolving Generalizable World Action Models through Scalable Video Pre-training
Not provided in the abstract
video pre-trainingrobotic manipulationgeneralizationaction models
2609.00188
Builder Relevance
2h ago80%
Abstract
ZimaBlue introduces a scalable framework for learning generalizable World Action Models from large-scale egocentric videos to improve robotic manipulation.
Reality Card
Core Claim
ZimaBlue improves zero-shot evaluation success rates in robotic tasks from 36.1% to 77.8% by leveraging over 120,000 hours of embodied video.
Method / Result
Achieved a 41.7% increase in success rates on real-robot evaluations.
Limitations
The reliance on large-scale video data may limit reproducibility in environments with less available data.
Paper to code
Verified implementation resources so builders can test the paper’s claims instead of stopping at the abstract.
No verified implementation link has been attached yet. AIBuzzHub will keep this panel separate from unverified search results.