PromptForge
返回列表
AI应用RAG知识库检索优化向量数据库

RAG系统知识库质量评估与优化报告生成器

输入你的RAG系统配置和示例查询，自动生成检索质量评估报告，包括召回率分析、chunk策略建议和重排序优化方案。

2 浏览4/5/2026

You are a RAG System Quality Auditor. Analyze my RAG (Retrieval-Augmented Generation) setup and generate a comprehensive optimization report.

My Current Setup

  • Document types: [PDF/HTML/Markdown/etc.]
  • Chunking strategy: [fixed-size/semantic/recursive]
  • Chunk size: [N tokens], overlap: [M tokens]
  • Embedding model: [model name]
  • Vector DB: [Pinecone/Weaviate/Chroma/etc.]
  • Reranker: [yes/no, model if yes]
  • Top-K retrieval: [K]

Sample Queries That Perform Poorly

  1. [Query 1] → Expected answer: [X], Got: [Y]
  2. [Query 2] → Expected answer: [X], Got: [Y]

Please Generate

  1. Diagnosis: Root cause analysis for each failed query
  2. Chunking Optimization: Recommend chunk size, overlap, and strategy
  3. Retrieval Pipeline: Suggest hybrid search, query expansion, or HyDE
  4. Reranking Strategy: Whether to add/change reranker
  5. Evaluation Framework: RAGAS-compatible test cases
  6. Implementation Plan: Step-by-step migration path

Format as a structured report with severity ratings (Critical/High/Medium/Low).