https://haleuan.github.io/categories/%E7%A0%94%E7%A9%B6%E7%94%9F%E7%A7%91%E7%A0%94%E8%AE%B0%E5%BD%95/ Recent content in 研究生科研记录 on Haleuan's Blog Hugo zh-cn Thu, 17 Apr 2025 01:19:16 +0000 https://haleuan.github.io/post/paperdive-simpleiseffective/ Thu, 17 Apr 2025 01:19:16 +0000 https://haleuan.github.io/post/paperdive-simpleiseffective/ <h2 id="metadata">Metadata</h2> <p><strong>Title</strong>: <a href="https://arxiv.org/abs/2410.20724">Simple Is Effective: The Roles of Graphs and Large Language Models in Knowledge-Graph-Based Retrieval-Augmented Generation</a></p> <p><strong>Venue</strong>: ICLR 2025</p> <p><strong>Source Code</strong>: <a href="https://github.com/Graph-COM/SubgraphRAG">Github</a></p> <h2 id="tldr">TL;DR</h2> <h3 id="motivation">Motivation：</h3> <ol> <li>Retriever效率和准确性之间的平衡。</li> <li>知识的结构信息没有被充分利用</li> </ol> <hr> <h2 id="sections">Sections</h2> <h3 id="the-subgraphrag-framework">The SubgraphRAG Framework</h3> <h4 id="图的建立">图的建立</h4> <ul> <li></li> </ul> <blockquote> <p>Given a training set of question-answer pairs D, the subgraph retriever learning problem can be formulated as the following problem:</p> https://haleuan.github.io/post/note-of-autogen/ Wed, 04 Sep 2024 20:47:24 +0800 https://haleuan.github.io/post/note-of-autogen/ 学习AutoGen的一些简单笔记和大致框架