VAC Memory System is an open-source conversational memory framework for LLM agents that achieved 80.1% accuracy on the LoCoMo 2025 benchmark through a hybrid retrieval architecture combining MCA gating, FAISS semantic search, BM25 lexical search, and cross-encoder reranking.
原文翻译:
VAC记忆系统是一个开源的大语言模型智能体对话记忆框架,通过结合MCA门控、FAISS语义搜索、BM25词法搜索和交叉编码器重排序的混合检索架构,在LoCoMo 2025基准测试中实现了80.1%的准确率。
CodeGraphContext is a powerful MCP server and CLI toolkit that transforms local code repositories into queryable graph databases, enabling AI assistants and developers to analyze code relationships, dependencies, and structures through natural language queries or command-line tools.
原文翻译:
CodeGraphContext是一个强大的MCP服务器和CLI工具包,可将本地代码仓库转换为可查询的图数据库,使AI助手和开发者能够通过自然语言查询或命令行工具分析代码关系、依赖关系和结构。
GEO (Generative Engine Optimization) is the practice of optimizing content to increase citations and recommendations from AI assistants like ChatGPT, Gemini, and Claude. It involves auditing prompts, identifying citation gaps, and implementing fixes to improve visibility in AI-powered search results.
原文翻译:
GEO(生成式引擎优化)是一种优化内容以增加AI助手(如ChatGPT、Gemini和Claude)引用和推荐的实践。它涉及审核提示词、识别引用差距并实施修复,以提高在AI驱动搜索结果中的可见性。
Piragi is a powerful, zero-config RAG framework that supports multiple data sources, vector stores, and advanced retrieval techniques with built-in citations and auto-updates.
原文翻译:
Piragi是一个功能强大、零配置的RAG框架,支持多种数据源、向量存储和高级检索技术,内置引用和自动更新功能。
KTransformers is a flexible framework for optimizing large language model inference and fine-tuning through CPU-GPU heterogeneous computing, featuring two core modules: kt-kernel for high-performance inference and kt-sft for efficient fine-tuning.
原文翻译:
KTransformers是一个通过CPU-GPU异构计算优化大语言模型推理与微调的灵活框架,包含两个核心模块:kt-kernel用于高性能推理,kt-sft用于高效微调。