RAG systems
PulseAugur coverage of RAG systems — every cluster mentioning RAG systems across labs, papers, and developer communities, ranked by signal.
2 天有情绪数据
-
LLM 处理 Markdown 优于原始 HTML，减少令牌浪费
一篇近期文章强调，直接将原始 HTML 输入大型语言模型 (LLM) 会导致上下文窗口嘈杂和令牌使用效率低下。作者认为，LLM 对清晰的 Markdown 的理解远优于 HTML，因为 HTML 通常包含导航菜单、广告和样式包装器等无关元素。在摄取之前将 HTML 转换为 Markdown 可以大大减少令牌数量，改善语义分块，并提高 RAG 系统和 AI 代理的整体准确性和一致性。
-
VectorSmuggle attack hides data in AI embeddings; VectorPin offers defense
Researchers have identified a new steganographic attack vector called VectorSmuggle, which allows attackers to hide data within embeddings stored in vector databases used by RAG systems. This method exploits the lack of…
-
AI agents leverage foundation models for diverse tasks, focusing on tools and planning
Chip Huyen's latest post, adapted from her book "AI Engineering," explores the concept of intelligent agents, defining them as entities that perceive and act within an environment. These agents leverage the advanced cap…