Omni-Decision: Evidence-Ledger Planning for Omni-Modal Agents

类型：arxiv
标识：2607.11433
链接：https://arxiv.org/abs/2607.11433
主分类：multimodal
形态：method
TLDR：Omni-modal agents must seek evidence across video, audio, web pages, and computation to answer questions. Their main bottleneck is planning: noisy multimodal observations accumulate in conversation history and disrupt later decisions, while multimodal models have limited capacity for multi-step planning. Controlled backend replacements support this diagnosis: replacing the planner causes a much larger performance loss than replacing the perception backend. We present Omni-Decision, an omni-modal agent built on evidence-ledger planning: it replaces the growing dialogue history with an explicit
副分类：agent
待LLM分类：否
来源文件：
/inbox/tom/_candidates/2026-09-30-agent-rag-longcontext-candidates.json

2026-07-16

笔记 2026-09-24-0950 · flyP 双论文精读 · Realtime-Venus + GAE

2026-09-24

笔记 Tom 文献雷达 · AI Agent / RAG / 长上下文 · 2026-07-23T14:40

2026-07-23

笔记 Tom 文献雷达 · Agent + RAG + LongContext · 2026-08-29

2026-08-29

笔记 Tom 文献雷达 · Agent RAG LongContext · 2026-09-06 晚

2026-09-06

文档 Tom 评 flyP · 2026-09-08 AgentVista 多模态长程视觉 grounding 评测基准精读

2026-09-08

arXiv

2607.11433

TLDR

Omni-modal agents must seek evidence across video, audio, web pages, and computation to answer questions. Their main bottleneck is planning: noisy multimodal observations accumulate in conversation history and disrupt later decisions, while multimodal models have limited capacity for multi-step planning. Controlled backend replacements support this diagnosis: replacing the planner causes a much larger performance loss than replacing the perception backend. We present Omni-Decision, an omni-modal agent built on evidence-ledger planning: it replaces the growing dialogue history with an explicit

Omni-Decision: Evidence-Ledger Planning for Omni-Modal Agents

相关 · 按语义匹配