Material

Slides

Reading List:

Dynamic and Parametric Retrieval-Augmented Generation 

Section 1:

Standard RAG:

Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks 

Retrieval-Augmented Generation for Large Language Models: A Survey 

Sparse Retrieval

A Language Modeling Approach to Information Retrieval 

The Probabilistic Relevance Framework: BM25 and Beyond 

Dense Retrieval

Dense Passage Retrieval for Open-Domain Question Answering 

Optimizing Dense Retrieval Model Training with Hard Negatives 

Section 2: Dynamic RAG

In-Context Retrieval-Augmented Language Models 

Interleaving Retrieval with Chain-of-Thought Reasoning for Knowledge-Intensive Multi-Step Questions 

Active Retrieval Augmented Generation 

Mitigating Entity-Level Hallucination in Large Language Models 

DRAGIN: Dynamic Retrieval Augmented Generation based on the Information Needs of Large Language Models 

Decoupling Knowledge and Context: An Efficient and Effective Retrieval Augmented Generation Framework via Cross Attention 

SeaKR: Self-aware Knowledge Retrieval for Adaptive Retrieval Augmented Generation 

Self-RAG: Self-reflective Retrieval Augmented Generation 

Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning 

Search-o1: Agentic Search-Enhanced Large Reasoning Models 

R1-Searcher: Incentivizing the Search Capability in LLMs via Reinforcement Learning 

Section 3: Parametric RAG

Lost in the Middle: How Language Models Use Long Contexts 

Same Task, More Tokens: The Impact of Input Length on the Reasoning Performance of Large Language Models 

RankRAG: Unifying Context Ranking with Retrieval-Augmented Generation in LLMs 

HyperNetworks 

Hypencoder: Hypernetworks for Information Retrieval 

Parametric Retrieval Augmented Generation 

Dynamic Parametric Retrieval Augmented Generation for Test-time Knowledge Enhancement 

Plug-in Parameter Generation for Test-time Parametric Knowledge Injection 

Toolbox

• Dynamic RAG 

• Parametric RAG