{"slug": "wiki-foundation-model-for-complex-agentic-reasoning", "title": "Wiki Foundation Model for Complex Agentic Reasoning", "summary": "Researchers Junnan Dong, Linhao Luo, Senlei Zhang and colleagues at Tencent Youtu Lab and Monash University proposed WFM, a Wiki Foundation Model for encoding and retrieving LLM Wiki knowledge bases that pair dense markdown documents with explicit links. WFM combines a Wiki Graph schema, query-conditioned attentive aggregation with attention-variance regularization, and an NCCL boundary exchange protocol using fixed-shape GPU-to-GPU collectives that bypasses CPU serialization, yielding 10.5x faster training. The team reported strong results across five long-term agent memory and multi-hop reasoning benchmarks.", "body_md": "# WFM: Wiki Foundation Model for Complex Agentic Reasoning\n\nJunnan Dong, Linhao Luo, Senlei Zhang and colleagues at Tencent Youtu Lab and Monash University propose WFM, a Wiki Foundation Model for encoding and retrieving LLM Wiki knowledge bases that pair dense markdown documents with explicit links.\n\n## Ask this paper\n\nWiki Graph schema. The schema keeps explicit link topology alongside dense text semantics.\n\nQuery-conditioned aggregation. Message passing uses query-conditioned attention with an attention-variance regularizer.\n\nDistributed training. An NCCL boundary exchange protocol with fixed-shape GPU-to-GPU collectives avoids CPU serialization and gives 10.5x faster training, with strong results on five agent memory and multi-hop benchmarks.\n\n## Abstract\n\nReal-world agents fundamentally require persistent non-parametric knowledge for dynamic reasoning, i.e., long-term memory and retrieval-augmented generation. While graphs have shown reliable advantages in providing structured evidence, the sparse graph representations naturally restrict machine readability and semantic density required for complex agentic workflows. Driven by this limitation, the entire industry is witnessing a paradigm shift from traditional sparse graphs to LLM Wiki, an agent-native knowledge representation that couples dense document contexts with markdown files containing multi-layered topological linkages. However, parameterizing such rich semantics is challenging to encode dense textual contexts using traditional sparse graph embeddings. Moreover, learning LLM Wiki with existing graph encoders could overwhelm distributed system overheads that hinder deployment in large-scale commercial scenarios. To this end, we propose a novel paradigm Wiki Foundation Model, i.e., WFM, tailored for scalable, agent-native representation and retrieval. Specifically, (i) we formalize a Wiki Graph schema that seamlessly bridges fine-grained structures with dense contexts, maintaining explicit topologies alongside continuous semantics; (ii) A query-conditioned attentive aggregation is tailored for rich wiki message passing and explicit attention variance regularization; (iii) We engineer an infrastructural NCCL boundary exchange protocol that hoists static partition indices and leverages fixed-shape GPU-to-GPU collectives, bypassing CPU serialization and memory copy overheads. Extensive evaluations across five long-term agent memory and multi-hop reasoning benchmarks demonstrate the remarkable performance of WFM, while achieving a 10.5 times training acceleration on distributed clusters.", "url": "https://wpnews.pro/news/wiki-foundation-model-for-complex-agentic-reasoning", "canonical_source": "https://academy.dair.ai/papers/wfm-wiki-foundation-model-for-complex-agentic-reasoning-2609.18182", "published_at": "2026-09-23 02:01:17+00:00", "updated_at": "2026-09-23 02:23:52.565546+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-agents", "ai-research", "large-language-models", "ai-infrastructure"], "entities": ["Tencent Youtu Lab", "Monash University", "WFM", "Junnan Dong", "Linhao Luo", "Senlei Zhang", "NCCL", "LLM Wiki"], "alternates": {"html": "https://wpnews.pro/news/wiki-foundation-model-for-complex-agentic-reasoning", "markdown": "https://wpnews.pro/news/wiki-foundation-model-for-complex-agentic-reasoning.md", "text": "https://wpnews.pro/news/wiki-foundation-model-for-complex-agentic-reasoning.txt", "jsonld": "https://wpnews.pro/news/wiki-foundation-model-for-complex-agentic-reasoning.jsonld"}}