LLM RAG

Configuration

Environment

• Copy .env.example file to .env
• Fill in the required values

Installation

Using pip

• Create and activate new Python virtual environment
• pip install -r requirements.txt
• pip install -e . (basic packages)
  or
  pip install -e .[ragatouille] (with ragatouille for ColBERT)

Using uv

• uv sync --group dev (basic packages)
  or
  uv sync --group dev --extra ragatouille (with ragatouille for ColBERT)
• uv pip install -e .

RAG From Scratch

YouTube Playlist
Repository

Part	Name	Video	Slides	Jupyter Notebook	Python Script	LangGraph Studio
1	Overview	Watch	View	01-overview.ipynb	-	-
2	Indexing	Watch	View	02-indexing.ipynb	-	-
3	Retrieval	Watch	View	03-retrieval.ipynb	-	-
4	Generation	Watch	View	04-generation.ipynb	-	-
5	Query Translation - Multi-Query	Watch	View	05-multi-query.ipynb	multi_query.py	Query Translation - Multi-Query
6	Query Translation - RAG-Fusion	Watch	View	06-rag-fusion.ipynb	rag_fusion.py	Query Translation - RAG-Fusion
7	Query Translation - Decomposition	Watch	View	07-01-decomposition-recursive.ipynb 07-02-decomposition-parallel.ipynb	recursive.py parallel.py	Query Translation - Decomposition (Recursive) Query Translation - Decomposition (Parallel)
8	Query Translation - Step-Back Prompting	Watch	View	08-step-back.ipynb	step_back.py	Query Translation - Step-Back Prompting
9	Query Translation - HyDE	Watch	View	09-hyde.ipynb	hyde.py	Query Translation - HyDE
10	Routing	Watch	View	10-01-logical-routing.ipynb	logical.py semantic.py	Routing - Logical Routing Routing - Semantic Routing
11	Query Construction	Watch	View	11-query-construction.ipynb	self_query.py	Query Construction - Self-Query
12	Indexing - Multi-Representation Indexing	Watch	View	12-01-multi-vector-summary.ipynb 12-02-multi-vector-chunks.ipynb 12-03-multi-vector-hypothetical-questions.ipynb	summary.py chunks.py hypothetical_questions.py	Indexing - Multi-Vector - Summary Indexing - Multi-Vector - Chunks Indexing - Multi-Vector - Hypothetical Questions
13	Indexing - RAPTOR	Watch	View	13-raptor.ipynb	raptor.py	-
14	Indexing - ColBERT	Watch	View	14-colbert.ipynb	colbert_model.py	-

Query Translation

RAG Fusion

Forget RAG, the Future is RAG-Fusion
RAG-Fusion: The Next Frontier of Search Technology
Reciprocal Rank Fusion outperforms Condorcet and individual Rank Learning Methods
Implementing Reciprocal Rank Fusion (RRF) in Python

Decomposition (Recursive)

Least-to-Most Prompting Enables Complex Reasoning in Large Language Models
Interleaving Retrieval with Chain-of-Thought Reasoning for Knowledge-Intensive Multi-Step Questions

Step-Back Prompting

Take a Step Back: Evoking Reasoning via Abstraction in Large Language Models

HyDE

Precise Zero-Shot Dense Retrieval without Relevance Labels

Routing

Semantic Router

Indexing

Multi-Representation Indexing

Dense X Retrieval: What Retrieval Granularity Should We Use?

RAPTOR

RAPTOR: Recursive Abstractive Processing for Tre-Organized Retrieval Building long context RAG with RAPTOR from scratch

ColBERT

ColBERT: Efficient and Effective Passage Search via Contextualized Late Interaction over BERT
ColBERTv2: Effective and Efficient Retrieval via Lightweight Late Interaction
RAGatouille
[Paper review] ColBERT, ColBERTv2
Overcoming the Limits of RAG with ColBERT
ColBERT Inference in the Browser