论文HF Daily Papers · arXiv 2611.04870昨天

PrimeSim：跨22平台3万小时数据训练的通用具身世界模型模拟器

PrimeSim基于30110小时跨22平台机器人演示数据训练，采用渐进式中后训练与双路径条件机制，实现10.1 FPS自回归仿真。在EWMBench评测领先，支持GRPO策略优化，已开源模型权重与训练代码。

AI 整理

仿真与训练
数据与采集
具身模型

本文由 AI 根据公开内容整理，可能有误，请以原文为准：PrimeSim: A Generalist World Model Simulator from Large-Scale Cross-Embodiment Demonstrations（HF Daily Papers · arXiv 2611.04870）

核心内容

PrimeSim是针对具身智能长程仿真痛点推出的动作条件化多视角世界模型。研究团队基于预训练视频生成器，使用涵盖22个机器人平台的30110小时演示数据进行渐进式中训练（Midtraining）。该模型解决了传统视频生成器在指令遵循、多视角一致性及长时序逻辑连贯性上的不足，推理速度达10.1 FPS，并在刚性及柔性操作任务中保持长程稳定性，相关模型权重、推理及训练代码均已开源。

技术上引入统一跨具身动作表示与帧对齐双路径条件机制，强化动作-帧对应关系；开发视图感知交错注意力模块，通过显式相机标识实现跨视角信息交换。后训练阶段采用“快视频-慢记忆”框架，结合因果生成、自回滚训练及分层时序记忆，将双向生成器转化为高效自回归模拟器。实测仅需100至200梯度步即可迁移至未见平台，在WorldSimProbe四项子集及EWMBench总分领先，且支持GRPO策略优化。

关键看点

大规模跨体数据：中训练阶段使用30110小时机器人演示数据，覆盖22个不同硬件平台，构建通用跨具身动作表征体系
架构创新：采用帧对齐双路径条件与视图感知交错注意力机制，配合快视频慢记忆后训练框架，解决长程误差累积问题
性能与迁移：仿真帧率10.1 FPS，在未见平台仅需100-200梯度步完成适配，EWMBench综合评分优于五个开源基线模型
策略优化支持：生成的长程Rollout具备足够保真度以支持GRPO算法进行策略优化，细粒度局部校准仍有提升空间

工程与产业观察

PrimeSim验证了利用海量异构机器人数据通过渐进式训练构建通用世界模型的可行性。其开源的3万小时跨平台数据集与“快视频-慢记忆”架构，为缓解具身仿真数据匮乏及Sim-to-Real鸿沟提供了关键基础设施，特别是支持GRPO优化的特性，使其成为当前端到端策略训练的重要基准与工具。

英文摘要 (Abstract)

A robot world model that serves as a simulator must do more than generate plausible videos: it must faithfully follow commanded actions, maintain consistency across camera views, and sustain coherent long-horizon rollouts. We present PrimeSim, an action-conditioned, multiview world model simulator that addresses these challenges through progressive midtraining and posttraining. Starting from a pretrained video generator, we midtrain on 30,110 hours of robot demonstrations spanning 22 platforms. To improve action following, we introduce a unified cross-embodiment action representation and frame-aligned dual-path conditioning to strengthen action-frame correspondence. To maintain multiview consistency, we develop view-aware interleaved attention with explicit camera identities, enabling cross-view information exchange while preserving view-specific representations. We further introduce a fast-video, slow-memory posttraining framework that transforms the bidirectional video generator into an efficient autoregressive simulator. By combining fast causal generation, self-rollout training, and hierarchical temporal memory, PrimeSim mitigates error accumulation from self-generated histories and supports coherent long-horizon simulation. PrimeSim runs at 10.1 FPS, remains stable over long-horizon rollouts on rigid and deformable manipulation tasks, and transfers to held-out platforms within 100 to 200 gradient steps where the pretrained backbone does not. Against five open-source action-conditioned world models it leads four of the five WorldSimProbe suites and posts the best overall EWMBench score, while fine-grained local calibration still trails the field; its rollouts are faithful enough to support GRPO-based policy optimization. We release the model weights, inference code, and training code.

Hugging Face Daily Papers 热度：9 upvotes 作者：Travis Davies, Wentao Ma, Lihao Guo, Tengfei Ma, Xiaowen Chang, Jiafeng Xu, Shilong Wang, Yiqi Deng, Shaowen He, Xiaoqi Li, Qian Zhu, Ran Cheng, Hao Dong