Advanced Retrieval and Re-ranking¶
A curated collection of research papers on dense retrieval, negative sampling strategies, and re-ranking techniques for information retrieval and question answering systems.
Contents:
- Overview of RAG and the Two-Stage Pipeline
- Building RAG Pipelines: A Practical Guide
- Computational Complexity Analysis of RAG Systems
- Overview
- The Hard Negative Problem
- Expert Perspectives: Architectures, Applications, and Trade-offs
- Benchmarks and Datasets for Retrieval and Re-ranking
- Comprehensive Comparison of Retrieval, Reranking, and RAG Libraries
- Introduction
- Taxonomy of Retrieval and Reranking Systems
- Full-Stack RAG Systems
- Research & Benchmarking Toolkits
- Reranking-Focused Libraries
- Vector Databases & Search Engines
- Retrieval-Specialized Libraries
- Detailed Comparison: Rankify vs Rerankers
- Performance Benchmarks
- Selection Guide
- Future Trends
- References
- Repository Links
- Literature Overview
- Stage 1: Retrieval Methods
- Sparse Retrieval Methods
- Dense Baselines & Fixed Embeddings
- Hard Negative Mining
- Late Interaction (ColBERT)
- Hybrid Dense-Sparse Methods
- Pre-training Methods for Dense Retrievers
- Joint Learning of Retrieval and Indexing
- Literature Survey: Retrieval Methods
- Overview
- Evolution of Stage 1 Methods
- Key Dimensions
- Quick Navigation
- The Central Challenge: Hard Negative Mining
- Stage 2: Re-ranking Methods
- Contributing
Welcome to the documentation for advanced retrieval and re-ranking research. This resource provides a comprehensive overview of the latest techniques and papers organized by their role in the RAG pipeline: Stage 1 (Retrieval) for efficient candidate selection and Stage 2 (Re-ranking) for precision scoring.
Quick Links¶
Overview of RAG and the Two-Stage Pipeline - Understanding RAG and the two-stage pipeline
Building RAG Pipelines: A Practical Guide - Building RAG pipelines: MVP to production
Computational Complexity Analysis of RAG Systems - Computational complexity analysis of RAG components
Expert Perspectives: Architectures, Applications, and Trade-offs - Expert opinions on architectures and trade-offs
Benchmarks and Datasets for Retrieval and Re-ranking - Evaluation benchmarks, datasets, and metrics
Comprehensive Comparison of Retrieval, Reranking, and RAG Libraries - 30+ libraries compared
Stage 1: Retrieval Methods - Stage 1: Dense retrieval methods
Stage 2: Re-ranking Methods - Stage 2: Re-ranking and cross-encoders
Contributing - How to contribute to this collection