arXiv：cs.AI· Wenlin Lin, Fanzhe Wei·· 13 小时前
KnowForge：面向开放域多模态文档问答的角色分工工具智能体与知识点索引
Locate, Judge, Read: Role-Specialized Tool Agents over a Knowledge-Point Index for Open-Domain Multimodal Document QA
AI 导读
KnowForge 用知识索引加 Locator、Deep Reader 等角色分工智能体，在完整 M3DocVQA（2,441 题、3,368 个 PDF）上以自托管 Qwen3.6-35B-A3B 达到 F1 47.5，较已发表的 M3DocRAG 高 11.0，用 DeepSeek-V4-Flash 则达 53.5。
正文
Abstract:Question answering over large heterogeneous document collections requires first finding the right passages and then reading them, yet current retrieval-augmented agents give one model both kinds of tools and let it schedule itself. We present KnowForge, a pipeline separated by job: a hybrid-retrieval knowledge-point index; a Locator agent that can only search; a code-scheduled bridge judge that forces the missing second hop; a typed, size-bounded evidence pool that merges every search's ranking; a Deep Reader agent that can only read text, images and tables; and a synthesizer with a format checker. With a self-hosted Qwen3.6-35B-A3B (3B active), KnowForge reaches F1 47.5 on the full open-domain M3DocVQA (2,441 questions, 3,368 PDFs), +11.0 over the published M3DocRAG system and +7.1 over one-shot retrieve-and-answer on the same index and model, and 53.5 with DeepSeek-V4-Flash. On MultiHop-RAG it reaches 73.5 accuracy, above the strongest published RAG and GraphRAG results (71.2); on the single-document MMLongBench-Doc it lifts the same model from 34.8 to 47.0, level with GPT-4o reading page images (46.3). Per-job ablations on the full benchmarks attribute the gain: the Deep Reader is worth 3.8, 9.8 and 0 points on the three corpora, tracking how visual each is; merging both agents into one with all tools and twice the rounds loses 1.9 F1 because it keeps searching instead of reading; a round-budget sweep shows a step at four Locator rounds and a plateau beyond; query latency scales with model calls rather than corpus size. We serve the three indexed benchmarks as read-only knowledge bases (free registration) and release all prompts and the runner code: this https URL
| Comments: | 17 pages, 3 figures, 11 tables. Code, prompts and benchmark runner: this https URL |
| Subjects: | Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Information Retrieval (cs.IR) |
| Cite as: | arXiv:2610.13128 [cs.AI] |
| (or arXiv:2610.13128v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.13128 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Wenlin Lin [view email]
[v1]
Fri, 9 Oct 2026 07:21:36 UTC (47 KB)
来源：arXiv：cs.AI · arxiv.org