Researchers have introduced RynnValue, an open-source value foundation model for robotic manipulation that utilizes temporal distance as a supervision target. This approach allows the model to scale to over 7,000 hours of data without requiring preference or progress annotations. RynnValue achieved a Kendall's tau_a of 0.675 on the RBM-EVAL-OOD benchmark, outperforming existing state-of-the-art methods. When converted into dense rewards, RynnValue significantly improved real-world policy success rates, demonstrating the effectiveness of temporal distance for generalist robot policies. AI
IMPACT Establishes temporal distance as a scalable supervision target for generalist robot policies, potentially accelerating progress in robot learning.
RANK_REASON The cluster describes a new research paper detailing a novel model and methodology for robotic learning.
- Alibaba DAMO Academy
- arXiv
- GitHub
- Hugging Face
- Kendall's tau_a
- modelscope
- RBM-EVAL-OOD
- Robotic Value Foundation Models
- RynnValue
- Temporal distance and discrimination: an audit study in academia
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
