APM-Bench: Benchmarking Cross-session Persistent Memory for Egocentric Streaming Video Assistants
- 类型：arxiv
- 标识：2609.37559
- 链接：https://arxiv.org/abs/2609.37559
- 主分类：multimodal
- 形态：benchmark
- TLDR：To serve as real-world personal assistants, streaming video models need persistent memory that retains past experiences for later use. Yet existing streaming benchmarks and methods often focus on individual continuous videos or short clips, overlooking that real-world interactions are often intermittent and require memory to persist across interruptions. To fill this gap, we introduce APM-Bench, which reformulates real-world streaming interaction as multi-session life trajectories. It contains 549 sessions, 104 trajectories, and 2,719 candidates, spanning both objective and open-ended question
- 副分类：evaluation
- 待LLM分类：否
- 标题中文：[标题中文] APM-Bench：面向自我中心流式视频助手的跨会话持久记忆基准
- TLDR中文：[TLDR中文] 要充当现实世界中的个人助手，流式视频模型需要持久记忆以保留过往经验供后续使用。然而现有流式基准与方法通常聚焦于单条连续视频或短视频片段，忽略了现实交互往往是间歇性的，且记忆需要在中断之间持续保持。为填补这一空白，我们提出 APM-Bench，将现实世界流式交互重构为多会话人生轨迹。该基准包含 549 个会话、104 条轨迹和 2,719 个候选，涵盖客观与开放式
- 来源文件：
- /inbox/tom/_candidates/2026-09-30-agent-rag-longcontext-candidates.json