LeRobot 发布 π₀.₅ LIBERO 微调 VLA 模型
原标题：lerobot/pi05_libero_finetuned_v044
AI 摘要
Hugging Face 上发布了 lerobot/pi05_libero_finetuned_v044，这是 Physical Intelligence 的 π₀.₅ 视觉-语言-动作（VLA）模型在 LIBERO 任务上微调后的 LeRobot 版本。该模型基于 flow matching 动作头，支持多视角图像、本体状态和可选语言指令输入，输出连续动作，可用于在真实环境中执行长时程任务。目前仅支持 flow-matching 动作头，原工作中的子任务预测、动作 tokenization 和 RL 等组件尚未开源。
正文节选
--- language: - en library_name: lerobot pipeline_tag: robotics tags: - vision-language-action - imitation-learning - lerobot inference: false datasets: - HuggingFaceVLA/libero license: gemma base_model: lerobot/pi05_libero_base --- # π₀.₅ (Pi05 Libero) (LeRobot) π₀.₅ is a Vision-Language-Action (VLA) model with open-world generalization from Physical Intelligence, co-trained on robot demonstrations and large-scale multimodal data to execute long-horizon tasks in unseen real-world environments