Physical Intelligence 推出多尺度具身记忆 MEM
To this end, we developed Multi-Scale Embodied Memory (MEM). The key idea: use different modalities to represent memory at different time scales.
AI 导读
为此，我们开发了多尺度具身记忆（MEM）。核心思路：用不同模态表示不同时间尺度上的记忆。 📹 对于短时程记忆，我们开发了一个高效的视频编码器，让模型能记住近期交互中的细粒度细节。 📜 对于长时程记忆，我们训练模型用文本总结事件，使其能记住长达 15 分钟的事件。
正文
当前显示原文，中文全文译文尚未就绪。
To this end, we developed Multi-Scale Embodied Memory (MEM). The key idea: use different modalities to represent memory at different time scales.
📹 For short horizon memory, we developed an efficient video encoder that lets the model remember fine-grained details about its recent interactions.
📜 For long horizon memory, we train the model to summarize events in text, allowing it to remember events for up to 15 min.