{"type":"rich","version":"1.0","provider_name":"Transistor","provider_url":"https://transistor.fm","author_name":"Embodied AI 101","title":"Video-Action Models for Robot Learning","html":"<iframe width=\"100%\" height=\"180\" frameborder=\"no\" scrolling=\"no\" seamless src=\"https://share.transistor.fm/e/f6ef801e\"></iframe>","width":"100%","height":180,"duration":1010,"description":"Introduces Video-Action Models (VAMs) that leverage pretrained internet-scale video models such as Cosmos-Predict2 as backbones instead of VLMs, paired with a flow-matching action decoder. Claims approximately 10x sample efficiency gains over standard vision-language-action models on real-world pick-and-place tasks.","thumbnail_url":"https://img.transistorcdn.com/l8CFsmXH35eVfcacIHndPwz_TJFZ0DzYC1nXc9Riruc/rs:fill:0:0:1/w:400/h:400/q:60/mb:500000/aHR0cHM6Ly9pbWct/dXBsb2FkLXByb2R1/Y3Rpb24udHJhbnNp/c3Rvci5mbS8wOGM3/YThiZDUxOTM4M2Vi/N2YzMTNkZDFiNDJh/ZDI1Mi5qcGc.webp","thumbnail_width":300,"thumbnail_height":300}