Week 7: Advanced RAG

Self-RAG, CRAG, RAPTOR, agentic retrieval

Week 7 of 12

Learning Objectives

  • Implement Self-RAG with reflection
  • Build corrective retrieval systems
  • Compare RAG strategies

Topics Covered

  • Self-RAG architecture
  • Corrective RAG (CRAG)
  • RAPTOR hierarchical retrieval
  • Adaptive retrieval strategies

Resources

Jupyter Notebooks

Open Self-RAG Implementation in Colab Self-RAG Implementation

Required Readings

PaperAuthorsYearLink
Self-RAG: Learning to Retrieve, Generate, and Critique Asai et al. 2023 arXiv
Corrective Retrieval Augmented Generation Yan et al. 2024 arXiv

Reading Guide: Advanced RAG

3-4 hours Self-RAG CRAG Adaptive retrieval

Study of Self-RAG, CRAG, and adaptive retrieval strategies

Primary Paper

Self-RAG: Learning to Retrieve, Generate, and Critique
Asai, A., Wu, Z., Wang, Y., et al. (2023)
arXiv arXiv

Secondary Papers

  • Corrective Retrieval Augmented Generation (CRAG) - Yan, S., et al. (2024) arXiv
  • RAPTOR: Recursive Abstractive Processing - Sarthi, P., et al. (2024) arXiv

Exercise: RAG Pipeline

100 Points 5-7 hours Advanced

Implement advanced RAG strategies

Learning Objectives

  • Create: Implement Self-RAG with reflection
  • Apply: Build corrective retrieval systems
  • Analyze: Compare RAG strategies

Tasks

TaskPointsDescription
Self-RAG Implementation 40 Implement Self-RAG with reflection tokens
CRAG Pipeline 35 Build corrective retrieval pipeline
Strategy Comparison 25 Benchmark RAG approaches

Exercise

Build an adaptive RAG system that:

  1. Decides when retrieval is needed
  2. Evaluates retrieval quality
  3. Corrects poor retrievals
  4. Generates grounded responses

Discussion Questions

  1. How does Self-RAG decide when to retrieve?
  2. What makes CRAG’s corrective approach effective?
  3. When should you use hierarchical vs. flat retrieval?

Additional Resources

Discussion & Questions

Join the Conversation

Have questions about this week's material? Want to discuss concepts with fellow students?



Back to top