Reproduce “RynnValue: Scaling Robotic Value Foundation Models with Temporal Distance” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Dongchi Huang, Hongyin Zhang, Bohan Hou, Siteng Huang, Zhian Su, Hang Guo, Tong Lu, Jiahao Tang and 6 more
RynnValue estimates how much time remains before an instructed goal, using timestamp-based labels from varied robot videos and training controls intended to prevent predictions from following sequence order. On an out-of-distribution benchmark, RynnValue-8B reaches average Kendall’s tau-a of 0.704, above Robometer’s 0.655. In four-task real-world tests, it also has higher success rates, although drawer placement remains challenging for both reward models.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.