Measuring Retrieval Robustness in LLMs, Training Small LLM Agents for End-to-End RAG, and More!

Vol.106 for May 26 - Jun 01, 2025

Stay Ahead of the Curve with the Latest Advancements and Discoveries in Information Retrieval.

This week’s newsletter highlights the following research:

1. Distilling Agentic Behavior into Compact Language Models, from Kang et al.

2. Training Small LLM Agents for End-to-End Retrieval-Augmented Generation, from Viettel Group

3. Self-Evolving Search Agents for Complex Question Answering, from Alibaba

4. Measuring and Understanding Retrieval Robustness in LLMs, from Bloomberg

5. Unifying Online Ad Ranking with Single-Model Architecture, from Meituan

6. Optimizing Query Decomposition for Multi-Vector Retrieval via LLM-Based Prompt Engineering, from Liu et al.

7. A Unified Framework for Hard Negative Mining in Enterprise Knowledge Retrieval, from Oracle AI

8. A Theoretical Analysis of Locality and Entropy in Neural Ranking, from The University of Glasgow

9. A Multi-Dataset Analysis of Chunk Size Effects in Dense Retrieval Systems, from Fraunhofer IAIS

10. A Systematic Framework for Understanding LLM-Based Recommendation Approaches, from NTU