Core Viewpoint - The listing of Zhipu marks a significant milestone for China's AI industry, being recognized as the first stock of a large model in the country, reflecting the culmination of three years of anticipation within the sector [2][15]. Company Background - Zhipu was founded in June 2019 by a team from Tsinghua University, focusing on natural language processing (NLP) and knowledge graphs, with initial funding of 40 million RMB [3]. - The company made a strategic decision to develop its own architecture, creating the General Language Model (GLM) instead of following the prevalent GPT and BERT paths [3][5]. Product Development - In August 2022, Zhipu open-sourced the GLM-130B model, which was recognized as the only Asian model in a major evaluation by Stanford University [5]. - Following the global success of ChatGPT, Zhipu released ChatGLM on March 14, 2023, alongside the open-sourcing of ChatGLM-6B, which became a popular tool for developers in China [6]. Market Position and Growth - By August 2023, Zhipu had entered a rapid growth phase, with model iterations occurring every three months and securing over 2.5 billion RMB in funding within the year [6][13]. - The GLM model was already operational in 12,000 enterprises globally, including nine of the top ten internet companies in China, as well as sectors like government, healthcare, and education [10]. Financial Performance - Despite rising revenues, Zhipu faces increasing R&D expenditures and significant losses, a common scenario among independent large model companies in Silicon Valley [13]. - The company’s financing before its IPO was substantial, with existing investors increasing their stakes and new strategic investors joining, indicating strong market confidence [13]. Future Outlook - The listing of Zhipu and its subsequent market performance will be closely scrutinized, as the company must navigate the challenges of maintaining growth and profitability in a competitive landscape [15][17].

侯明才同志长期从事沉积地质学与资源能源勘探相关的科研与教学工作。致力沉积学科数智化转型，研发了基于机器学习和知识图谱的沉积相识别与有机 碳预测方法，推动数智古地理重建平台创新，重建了东特提斯二叠纪以来高精度古地理演化模型；聚焦深层油气勘探，发展了面向沉积型资源能源勘探的 层序岩相古地理学，提出了重大构造古地理变革控源储物质和大型油气聚集带理论，支撑多个大型深层油气田勘探发现；瞄准绿色低碳能源发展需求，提 出青藏高原及东缘"原—山—盆"差异控热机制，牵头组建"产—学—研—用"一体化团队，推进地热资源评价与综合利用示范。 在NC、Geology、地质学报等期刊发表论文300余篇，出版专著10部，获国家科技进步奖二等奖1项（R04）、省部级一等奖1项（R03）、国土资源科学技 术奖二等奖1项（R05）。为四川省学术和技术带头人、四川省有突出贡献优秀专家、天府万人计划创新领军人才，享受国务院政府特殊津贴。 侯明才，男，1968年10月生，二级教授，中共党员，四川省南部县人。1992年毕业于东华理工大学（原华东地质学院）铀矿地质勘查专业，2000、2003年 毕业于成都理工大学沉积地质学专业，先后获硕士和博士学位。20 ...

Technology & Software Development - Graph RAG is presented as a practical example for RAG over code, addressing limitations of naive chunking in handling codebases with long-range dependencies [1] - Graph-Code, a graph-driven RAG system, is introduced for analyzing Python codebases and enabling natural language querying [1] - Graph-Code extracts classes, functions, and relationships from code through deep code parsing [1] - Memgraph is utilized to store the codebase as a graph within the Graph-Code system [1] - Graph-Code parses pyproject files to understand external dependencies [1] - The system retrieves actual source code snippets for found functions [1]

Join the DKG Global Hackathon to take on challenges in the AI industry using blockchain + knowledge graph tech.$30K prize poolSubmission Deadline: Nov 21Register today: https://t.co/thkbJUBj77OriginTrail (@origin_trail):In the age of AI, how do we know what’s true?Do you have what it takes to build:•⁠ Grokpedia vs. Wikipedia comparison•⁠ Decentralized Community Notes•⁠ ⁠Reputation Social Graph💰$30K rewardsSponsored by @Polkadot, @umanitek & @origin_trailLink in reply. https://t.co/smoq51JtXQ ...

Agents forget everything after each task!Graphiti builds a temporal knowledge graph for Agents that provides a memory layer to all interactions.Fully open-source with 20k+ stars!Learn how to use Graphiti MCP to connect all AI apps via a common memory layer (100% local): https://t.co/cpAZFJcrufAvi Chawla (@_avichawla):Big update for Claude Desktop and Cursor users!Now you can connect all AI apps via a common memory layer in a minute.I used the Graphiti MCP server that runs 100% locally to cross-operate acros ...

Core Insights - The sales intelligence market is crowded, with services that help identify prospects, provide background information, and automate follow-ups [1] - Sumble, a startup from San Francisco, aims to provide contextual information by aggregating data from various online sources [2] Company Overview - Sumble was founded by Anthony Goldbloom and Ben Hamner, who previously created the data science community Kaggle [3] - The startup utilizes a knowledge graph supported by large language models to connect diverse data points, offering insights into a company's technographic data, organizational structure, and potential contacts [3] Market Position and Growth - Despite the competitive landscape, Sumble has successfully signed 17 enterprise customers since its launch in April 2024, including notable companies like Snowflake and Figma [4] - The startup has experienced significant growth, with a reported 550% year-over-year revenue increase, although specific revenue figures were not disclosed [4] User Engagement and Funding - Sumble's user base has grown rapidly within companies, often expanding from a few users to hundreds in a short period, primarily through word of mouth and internal communication channels like Slack [5] - The company recently emerged from stealth mode with $38.5 million in funding, including an $8.5 million seed round and a $30 million Series A led by prominent investors [5]

RT Avi Chawla (@_avichawla)Build human-like memory for your Agents (open-source)!Every agentic and RAG system struggles with real-time knowledge updates and fast data retrieval.Zep solves these issues with its continuously evolving and temporally-aware Knowledge Graph.Like humans, Zep organizes an Agent's memories into episodes, extracts entities and their relationships from these episodes, and stores them in a knowledge graph:(refer to the image below as you read)1) Episode Subgraph: Captures raw data with ...

Agentic System Challenges - Agentic 和 RAG 系统在实时知识更新和快速数据检索方面面临挑战 [1] Zep's Solution - Zep 通过其不断发展和时间感知的知识图谱来解决这些问题 [1] - Zep 像人类一样组织信息 [1]

Core Functionality - Zep aims to build human-like memory for agents, addressing real-time knowledge updates and fast data retrieval challenges in agentic and RAG systems [1] - Zep organizes agent memories into episodes, extracts entities and relationships, and stores them in a knowledge graph [1] - The system features an Episode Subgraph for capturing raw, timestamped data, a Semantic Entity Subgraph for extracting and versioning entities and facts, and a Community Subgraph for grouping related entities [1][2] Performance Metrics - Zep delivers up to 1850% (18.5 times) higher accuracy with 90% lower latency compared to tools like MemGPT [2] Open Source Nature - Zep is fully open-source [2]

Graph RAG Overview - Graph RAG aims to enhance LLMs by incorporating knowledge graphs, addressing limitations like lack of domain knowledge, unverifiable answers, hallucinations, and biases [1][3][4][5][9][10] - Graph RAG leverages knowledge graphs (collections of nodes, relationships, and properties) to provide more relevant, contextual, and explainable results compared to basic RAG systems using vector databases [8][9][10][12][13][14] - Microsoft research indicates Graph RAG can achieve better results with lower token costs, supported by studies showing improvements in capabilities and analyst trends [15][16] Knowledge Graph Construction - Knowledge graph construction involves structuring unstructured information, extracting entities and relationships, and enriching the graph with algorithms [19][20][21][22] - Lexical graphs represent documents and elements (chunks, sections, paragraphs) with relationships based on document structure, temporal sequence, and similarity [25][26] - Entity extraction utilizes LLMs with graph schemas to identify entities and relationships from text, potentially integrating with existing knowledge graphs or structured data like CRM systems [27][28][29][30] - Graph algorithms (clustering, link prediction, page rank) enrich the knowledge graph, enabling cross-document topic identification and summarization [20][30][34] Graph RAG Retrieval and Applications - Graph RAG retrieval involves initial index search (vector, full text, hybrid) followed by traversing relationships to fetch additional context, considering user context for tailored results [32][33] - Modern LLMs are increasingly trained on graph processing, allowing them to effectively utilize node-relationship-node patterns provided as context [34] - Tools and libraries are available for knowledge graph construction from various sources (PDFs, YouTube transcripts, web articles), with open-source options for implementation [35][36][39][43][45] - Agentic approaches in Graph RAG break down user questions into tasks, using domain-specific retrievers and tools in sequence or loops to generate comprehensive answers and visualizations [42][44] - Industry leaders are adopting Graph RAG for production applications, such as LinkedIn's customer support, which saw a 286% reduction in median per-issue resolution time [17][18]