Advanced RAG Techniques 🚀
Explore all the advanced RAG and Query Enhancement techniques for improving retrieval and generation.
Basic RAG
Understand the basics and fundamentals about RAG
Query Transformations
Enhance query formulation through transformations and sub-queries.
Hypothetical Questions (HyDE Approach)
Generate hypothetical questions to align queries better with data.
Contextual Chunk Headers
Add context-based headers to chunks for better retrieval accuracy.
Relevant Segment Extraction
Dynamically extract multi-chunk segments relevant to queries.
Context Enrichment Techniques
Enrich retrieval accuracy by providing additional contextual data.
Contextual Compression
Compress information while retaining relevant content.
Document Augmentation via Question Generation
Augment documents with questions for improved retrieval.
Fusion Retrieval
Combine different retrieval models for optimal results.
Intelligent Reranking
Advanced reranking to improve retrieval relevance.
Two-Stage Retrieval
Implement a two-step retrieval process for better accuracy.
Multi-faceted Filtering
Filter retrieved results using metadata, relevance scores, and diversity.
Hierarchical Indices
Organize information in a hierarchical index for efficient retrieval.
Ensemble Retrieval
Apply multiple retrieval methods and models for robust results.
Multi-modal Retrieval
Extend retrieval to handle multiple data types like images and videos.
Retrieval with Feedback Loops
Iteratively improve retrieval by incorporating user feedback.
Adaptive Retrieval
Dynamically adjust retrieval strategies based on query types.
Iterative Retrieval
Refine results by performing multiple rounds of retrieval.
Explainable Retrieval
Provide transparency in the retrieval process to build user trust.
Knowledge Graph Integration (Graph RAG)
Use knowledge graphs to enrich retrieved information.
RAPTOR: Recursive Abstractive Processing
Organize retrieved information in a tree structure for better context.
Self RAG
A dynamic method that combines both retrieval and generation-based approaches.
Corrective RAG
Dynamically evaluates and corrects the retrieval process.
Sophisticated Controllable Agent for Complex RAG Tasks
Uses advanced planning to answer complex questions that simple retrieval cannot solve.