# 记忆即权重：将长期历史内化用于流式视频

> 原文链接：https://robotics.hellobike.com/papers/2026-10-15-2610-22070
> 频道：papers | 更新日期：2026-10-15 | 标签：arxiv, cs.RO
> 核心摘要：MaLoW将历史上下文内化为轻量参数更新，用于流式视频理解，并在多个基准及VLA模型上验证。

# 记忆即权重：将长期历史内化用于流式视频

::: v-pre
> 原文链接：[Memory as Weights: Internalizing Long-Term History for Streaming Videos](https://arxiv.org/abs/2610.22070)（arXiv 2610.22070 · 发布日期：2026-10-15）

**Memory as Weights: Internalizing Long-Term History for Streaming Videos**

MaLoW将历史上下文内化为轻量参数更新，用于流式视频理解，并在多个基准及VLA模型上验证。

## 核心要点

- 现有记忆机制将历史重新放入输入上下文，与当前观察争夺有限上下文。
- MaLoW用循环HyperNetwork将视频观察增量整合为固定大小LoRA权重。
- 构建MaLoW-Instruct，含381K问答对和56K视频，支持流式视频训练。
- 在OVO-Bench backward split较既有记忆方法提升3.9%至11.8%。

**验证范围与局限**：验证范围包括OVO-Bench、StreamingBench、StreamArena和Simpler WidowX。

## 英文摘要 (Abstract)

Streaming vision-language models (VLMs) must continuously integrate information from the past while remaining responsive to current observations and user queries. Existing memory mechanisms typically compress, retrieve, or cache historical observations and reintroduce them into the input context, forcing historical memory and current observations to compete for limited context. We propose MaLoW, a streaming memory framework that instead internalizes historical context into evolving lightweight parameter updates. MaLoW uses a recurrent HyperNetwork to incrementally consolidate incoming video observations into fixed-size LoRA weights, allowing historical information to be carried forward without revisiting the full video stream. In doing so, MaLoW represents history as weights, while keeping the current observation as input. To support training, we further construct MaLoW-Instruct, a large-scale streaming video dataset containing 381K question-answer pairs over 56K videos. We conduct extensive experiments to demonstrate that MaLoW achieves strong performance on OVO-Bench, StreamingBench, and StreamArena. On the backward split of OVO-Bench, MaLoW outperforms prior memory approaches by 3.9% to 11.8% across different video backbones while incurring only modest latency overhead. Beyond streaming video understanding, MaLoW also generalizes to vision-language-action models, improving the action success rate on Simpler WidowX by 9.3%.

作者：Zhuoming Liu, Jinhong Lin, Lin Zhang, Yiwu Zhong, Yin Li
:::

## 原始参考来源 (Sources)

- [arXiv 2610.22070：Memory as Weights: Internalizing Long-Term History for Streaming Videos](https://arxiv.org/abs/2610.22070)