搜索结果：人工智能

RΞASON is an open-source TypeScript backend framework designed for building LLM applications, featuring structured output using TypeScript interfaces, functional agents, native streaming support, and zero-setup observability. 原文翻译： RΞASON是一个开源的TypeScript后端框架，专为构建LLM应用而设计，具有使用TypeScript接口的结构化输出、函数式代理、原生流式支持以及零配置可观测性。

This article explores an alternative approach to the Model Context Protocol (MCP) for integrating tools with Large Language Models (LLMs) by leveraging existing OpenAPI servers. It proposes a simpler, more intuitive method that uses structured HTTP API definitions as tool inputs, requiring only minimal authentication flow additions. The implementation is demonstrated through a concise Scala script, focusing on core tool integration while omitting MCP's broader features like prompts and resources. 原文翻译： 本文探讨了一种替代模型上下文协议（MCP）的方法，通过利用现有的OpenAPI服务器为大型语言模型（LLM）集成工具。它提出了一种更简单、更直观的方法，使用结构化的HTTP API定义作为工具输入，仅需添加最小的身份验证流程。通过一个简洁的Scala脚本演示了实现，专注于核心工具集成，同时省略了MCP更广泛的功能，如提示和资源。

Mastra implemented the LongMemEval benchmark to improve agent memory, achieving 80% accuracy through iterative optimizations including tailored templates, smarter memory updates, and better formatting - demonstrating RAG's continued effectiveness for agent memory. 原文翻译： Mastra通过实施LongMemEval基准测试来改进智能体记忆，通过定制模板、更智能的内存更新和更好的格式化等迭代优化，实现了80%的准确率，证明了RAG在智能体记忆中的持续有效性。

A local BM25 RAG pipeline for the Pi coding agent that indexes files and enables hybrid search with zero cloud dependency, working fully offline. 原文翻译： 一个用于Pi编码代理的本地BM25 RAG管道，可索引文件并实现混合搜索，无需云依赖，完全离线工作。

Building production-grade AI agents requires extensive optimization of both individual tools and end-to-end workflows, as accuracy compounds across multiple steps and small improvements in each component are critical for overall system reliability. 原文翻译：构建生产级AI智能体需要对单个工具和端到端工作流程进行广泛优化，因为准确性在多个步骤中会累积，每个组件的微小改进对于整体系统可靠性至关重要。

Prompt Refiner is a Python library for optimizing AI agent prompts, reducing API costs by 5-70% through token compression, cleaning, and smart context management. 原文翻译： Prompt Refiner 是一个用于优化AI Agent提示词的Python库，通过令牌压缩、清理和智能上下文管理，可将API成本降低5-70%。

Microsoft Research's GraphRAG enhances LLM capabilities by generating knowledge graphs from private datasets, significantly improving question-answering performance and enabling whole-dataset reasoning through graph machine learning. 原文翻译： 微软研究院的GraphRAG通过从私有数据集中生成知识图谱，显著提升大型语言模型的能力，通过图机器学习大幅改善问答性能并实现全数据集推理。

GraphRAG (Graphs + Retrieval Augmented Generation) is a technique for richly understanding text datasets by combining text extraction, network analysis, and LLM prompting and summarization into a single end-to-end system. 原文翻译： GraphRAG（图+检索增强生成）是一种通过将文本提取、网络分析、LLM提示和摘要整合到单一端到端系统中，来深入理解文本数据集的技术。

LLM Knowledge Graph Builder transforms unstructured data into Neo4j knowledge graphs, offering superior accuracy over traditional RAG through GraphRAG's relationship-aware retrieval mechanism. 原文翻译： LLM知识图谱构建器将非结构化数据转换为Neo4j知识图谱，通过GraphRAG的关系感知检索机制，提供比传统RAG更优越的准确性。

This research paper reveals new indirect prompt injection attack vectors that enable adversaries to remotely exploit LLM-integrated applications by injecting malicious prompts into retrieved data, demonstrating practical attacks against systems like Bing Chat and highlighting significant security vulnerabilities in current LLM deployments. 原文翻译： 本研究论文揭示了新的间接提示注入攻击向量，攻击者可通过在检索数据中注入恶意提示来远程利用LLM集成应用，展示了针对Bing Chat等系统的实际攻击，并突显了当前LLM部署中的重大安全漏洞。