{"version":1,"type":"story","url":"https://digestai.news/story/hugging-faces-jam-model-cuts-robotics-training-costs-by-73-without-vid","json":"https://digestai.news/story/hugging-faces-jam-model-cuts-robotics-training-costs-by-73-without-vid.json","markdown":"https://digestai.news/story/hugging-faces-jam-model-cuts-robotics-training-costs-by-73-without-vid.md","slug":"hugging-faces-jam-model-cuts-robotics-training-costs-by-73-without-vid","headline":"Hugging Face’s JAM model cuts robotics training costs by 73% without video generation","summary":"Hugging Face released a paper titled *Vision-Language-Action or World-Action? Just Action Models*, proposing a new approach called **JAM** that eliminates video generation during robotics training. The method uses a pretrained visual backbone and an action expert, trained via a *foresight loss* that predicts frozen future frame encodings—without retaining the generation branch at inference. 65.27%). The key innovation is a **73% reduction in FLOPs per sample** during training, measured on 8 NVIDIA GH200 GPUs, by caching foresight targets.\n\nThe paper argues the debate over vision-language-action (VLA) vs. world-action models (WAM) is overstated, as the performance gap isn’t tied to a single component—backbone initialization, future-observation objectives, or generation machinery. Cross-backbone tests on LIBERO-Plus show comparable results across Qwen3.5-2B (81.14%), Ovis-U1 (79.95%), and DINOv2-L (70.39%), all under JAM’s foresight recipe. The authors release public code built on FastWAM and ImageWAM, positioning JAM as a compute-efficient alternative to generative approaches in robotics.","keyPoints":["JAM cuts robotics training FLOPs by 73% vs. FastWAM by caching foresight targets on 8 GH200 GPUs","On LIBERO benchmark, JAM improves success rates from 97.0%/55.3%/91.2% to 98.2%/64.5%/91.4%","Paper claims vision-language backbones and generative models perform equally well under JAM’s foresight loss"],"whyItMatters":"JAM could redefine robotics training efficiency by removing video generation, slashing compute costs while matching or exceeding generative models’ performance. This challenges industry assumptions about pretrained vision-language models’ necessity and may accelerate deployment of AI-driven robots.","category":{"slug":"robotics","name":"Robotics & Physical AI","url":"https://digestai.news/category/robotics"},"entities":{"companies":["Hugging Face"],"models":["JAM","FastWAM","ImageWAM","Qwen3.5-2B","Ovis-U1","DINOv2-L"],"people":[]},"firstPublishedAt":"2026-10-27T21:58:00Z","updatedAt":"2026-10-27T21:58:00Z","sourceCount":1,"hasPrimarySource":false,"sources":[{"outlet":"aiweekly.co","title":"Hugging Face Paper: Robot Policies Don't Need Video Generation","url":"https://aiweekly.co/alerts/hugging-face-paper-robot-policies-dont-need-video-generation","publishedAt":"2026-10-27T21:58:00Z","type":"press","primary":false,"lead":true}],"sourceNotes":null,"discussions":[],"thread":{"title":"InclusionAI's Accelerating AI Model Saga","url":"https://digestai.news/thread/inclusionai-releases-ming-3-0-audio-flash-a-multilingual-audio-language-model","storyCount":4},"cite":{"text":"Digest AI, \"Hugging Face’s JAM model cuts robotics training costs by 73% without video generation\", 27 October 2026, https://digestai.news/story/hugging-faces-jam-model-cuts-robotics-training-costs-by-73-without-vid","publisher":"Digest AI","title":"Hugging Face’s JAM model cuts robotics training costs by 73% without video generation","datePublished":"2026-10-27T21:58:00Z","url":"https://digestai.news/story/hugging-faces-jam-model-cuts-robotics-training-costs-by-73-without-vid"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}