SJTU and Noematrix's SWAM Scales World-Action Model on 47K Hours of Video, 40% vs 23% Real-World Baseline
Summary
Shanghai Jiao Tong and Noematrix's SWAM investigates how language-video, language-video-action and joint supervision shape world-action-model performance using a curated 47,000-hour pool of general video, egocentric footage, UMI demos and robot data. SWAM hits 88.10% in sim clean-to-clean, 61.56% clean-to-random, and 40.00% real-world success vs 23.33% for the strongest baseline; datasets, code and weights are open.
Originally reported by arxiv.org
Read the original article →Original headline: SJTU and Noematrix's SWAM Scales World-Action Model on 47K Hours of Video, 40% vs 23% Real-World Baseline