Stanford Team Develops OpenWAM Framework for World-Action Models
Overview
Stanford's Li Fei-Fei team introduces OpenWAM, an open framework for world-action models that standardizes training and evaluation. It uses causal video pre-training, modular components, and a shared MoT architecture to improve task success rates across diverse robotic tasks. The framework emphasizes sim-to-real transfer and the use of diverse, counterfactual data for robust action models.
Generated from attributed reports · Updated 3 hours ago
Event evidence and corrections
1 attributed source owners. Ownership does not establish independent confirmation. Quantities are reported separately and are never added together.
Reported quantity · training days: 14 other · Basis not reported
“同一副骨架，四种「先想还是先动」 OpenWAM 从阿里开源的视频模型 Wan2.2-5B 出发，先在约 334 万条机器人与人类交互视频上继续预训练，名义时长约 1.46 万小时，全程不用动作标签，在 32 块 B200 上训练了 14 天。”
Exact source · revision 1jiqizhixin
“同一副骨架，四种「先想还是先动」 OpenWAM 从阿里开源的视频模型 Wan2.2-5B 出发，先在约 334 万条机器人与人类交互视频上继续预训练，名义时长约 1.46 万小时，全程不用动作标签，在 32 块 B200 上训练了 14 天。”
Exact source · revision 2Source owner not reported
Reported quantity · number of segments: 32000 other · Basis not reported
“为此团队构建了 LIBERO-Long-CF 数据集 ：恢复示范中的仿真器状态，执行被改动过的动作，比如停止、反向、加噪声、改变夹爪开合时机等，共得到 32000 个片段、约 410 万条控制记录，是原始示范的 29.7 倍，其中 72.3% 的片段改变了物体的摆放状态。”
Exact source · revision 1jiqizhixin
“为此团队构建了 LIBERO-Long-CF 数据集 ：恢复示范中的仿真器状态，执行被改动过的动作，比如停止、反向、加噪声、改变夹爪开合时机等，共得到 32000 个片段、约 410 万条控制记录，是原始示范的 29.7 倍，其中 72.3% 的片段改变了物体的摆放状态。”
Exact source · revision 2Source owner not reported
Reported quantity · Reported success rate: 92.1 percent · Basis not reported
“真机上，团队用两台 Franka FR3 机械臂测试了烤面包、还原 2×2 魔方最后一层和按颜色分拣杯子，VTA 和 Joint 平均成功率分别为 92.1% 和 91.9%。”
Exact source · revision 1jiqizhixin
“真机上，团队用两台 Franka FR3 机械臂测试了烤面包、还原 2×2 魔方最后一层和按颜色分拣杯子，VTA 和 Joint 平均成功率分别为 92.1% 和 91.9%。”
Exact source · revision 2Source owner not reported
Reported quantity · training hours: 14600 hours · Basis not reported
“同一副骨架，四种「先想还是先动」 OpenWAM 从阿里开源的视频模型 Wan2.2-5B 出发，先在约 334 万条机器人与人类交互视频上继续预训练，名义时长约 1.46 万小时，全程不用动作标签，在 32 块 B200 上训练了 14 天。”
Exact source · revision 2Source owner not reported
Report timeline
Follow attributed reports and material updates.
- Chinese robotics — Robot learning researchSignalOpenWAM: An Open Framework for Composable World-Action Models
This paper presents OpenWAM, an open framework for world-action models (WAM) that aims to standardize the evaluation of different WAM designs. The framework includes a three-stage training process, a shared MoT architecture, and four configurable interaction modes. It introduces modular components like IDM and FDM, trained on a large-scale 'counterfactual' dataset, leading to improved task success rates in both simulation and real-world robotic tasks.
- Machine Heart — Robotics on WeChatSignalLi Fei-Fei's Team Builds a World Action Model: Switching the Base, Robot Success Rate Soars Directly
Stanford's Li Fei-Fei team introduces OpenWAM, an open framework for world-action models (WAM) that standardizes training and evaluation. The framework uses causal video pre-training, modular components like IDM and FDM, and a shared MoT architecture to improve task success rates across diverse robotic tasks. The paper highlights the importance of sim-to-real transfer and the value of diverse, counterfactual data in training robust action models.
Timezone · UTC
Article dates follow your selected timezone. Briefing editions use Hong Kong time (UTC+8).