Reproduce “FLORA: An Open Model Suite for Scaling Robotics Data Engines” 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
Sicong Leng, Yunze Man, Zhiqi Li, Jing Wang, Yuhe Zhang, Ryan Punamiya, Siddharth Patki, Kaiyuan Zheng and 9 more
Flora turns robot videos into timestamped intended-action captions and execution-quality labels. It segments actions, describes them with nearby visual context, and uses one reward model for episode, step, or user-defined rubric scores. On a fixed 50-episode subset, aggregate quality was 57.1% versus 53.7% for maximum-effort GPT-6 Astra, at an estimated 125-times lower cost; this is a system-level, modeled-cost comparison.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.