ResearchBenchmarksSweep toys on Real-world HardwareFollow90Success RateiTuP6.828.45071.6May 2, 2025Evaluation ResultsMethodMethodLinksSuccess RateiTuP2025.0590iTuPSDG-Net=ExcludedSDG-Net=Excluded2025.0570CoPa2025.0570iTuP2-level granularity=Ex...2-level granularity=Excluded2025.0560iTuPVLM=ExcludedVLM=Excluded2025.0510