Generative Retrieval

Follow this topic Ask about this topic

Generative retrieval studies how to find documents, products, or other items by generating their identifiers directly from a query rather than ranking a precomputed candidate set. Many systems encode items as semantic IDs, then use autoregressive or diffusion models with constrained decoding; training can combine supervised examples, contrastive objectives, and reinforcement learning from relevance or user-preference signals. The approach is being applied to web and e-commerce search, recommendation, legal and multimodal retrieval, where it can simplify retrieval pipelines or incorporate richer context. Key challenges include preserving exact-match recall and identifier quality, controlling decoding cost and errors, and establishing fair comparisons with dense retrieval systems.

Latest papers

45 papers7 papers in the last four weeks, down 22% on the four weeks before. 0.0% of all new papers.

ORBIT: Preserving Foundational Language Capabilities in GenRetrieval via Origin-Regulated Merging
May 12, 2026Neha Verma, Nikhil Mehta, Shao-Chuan Wang +7Continual Learning for LLMs LLM Fine-Tuning
Johns Hopkins University · Google DeepMind · Google
How Generative AI Disrupts Search: An Empirical Study of Google Search, Gemini, and AI Overviews
Apr 30, 2026Riley Grossman, Songjiang Liu, Michael K. Chen +3Generative Engine Optimization Generative Retrieval
New Jersey Institute of Technology Newark, NJ, USA · Nanyang Technological University Singapore, Singapore · Indiana University Bloomington Bloomington, IN, USA
Lost in Decoding? Reproducing and Stress-Testing the Look-Ahead Prior in Generative Retrieval
Apr 25, 2026Kidist Amde Mekonnen, Yongkang Li, Yubao Tang +2Language Model Decoding Generative Retrieval
University of Amsterdam Amsterdam, The Netherlands
A Parametric Memory Head for Continual Generative Retrieval
Apr 25, 2026Kidist Amde Mekonnen, Yubao Tang, Maarten de RijkeMemory-Augmented Neural Networks LLM Memory
University of Amsterdam Amsterdam, The Netherlands
GenRec: A Preference-Oriented Generative Framework for Large-Scale Recommendation
Apr 16, 2026Yanyan Zou, Junbo Qi, Lunsong Huang +7RL for Recommendation Generative Retrieval
JD.com Beijing, China · Waseda University Tokyo, Japan
From Relevance to Authority: Authority-aware Generative Retrieval in Web Search Engines
Apr 15, 2026Sunkyung Lee, Jihye Back, Donghyeon Jeon +4Information Retrieval Generative Retrieval
Sungkyunkwan University, Republic of Korea · Naver Corporation, Republic of Korea
OneSearch-V2: The Latent Reasoning Enhanced Self-distillation Generative Search Framework
Mar 25, 2026Ben Chen, Siyuan Wang, Yufei Ma +20Generative Retrieval E-Commerce Search
Kuaishou Technology, Beijing, China
Deploying Semantic ID-based Generative Retrieval for Large-Scale Podcast Discovery at Spotify
Mar 18, 2026Edoardo D'Amico, Marco De Nadai, Praveen Chandar +41Generative Recommendation Large Language Model-Based Recommendation
CRE-T1 Preview Technical Report: Beyond Contrastive Learning for Reasoning-Intensive Retrieval
Mar 18, 2026Guangzhi Wang, Yinghao Jiao, Zhi LiuGenerative Retrieval Implicit Reasoning in Language Models
CareerInternational Research Team
Answer Bubbles: Information Exposure in AI-Mediated Search
Mar 17, 2026Michelle Huang, Agam Goyal, Koustuv Saha +1Generative Retrieval Information Retrieval
Siebel School of Computing and Data Science University of Illinois Urbana-Champaign, Urbana, IL, USA
Multi-Step Semantic Reasoning in Generative Retrieval
Mar 12, 2026Steven Dong, Yubao Tang, Maarten de RijkeGenerative Retrieval Numerical Reasoning in Language Models
University of Amsterdam, Amsterdam, The Netherlands
Aligning Large Language Models with Searcher Preferences
Mar 11, 2026Wei Wu, Peilun Zhou, Liyi Chen +6Personalized Language Model Alignment Generative Retrieval
School of Artificial Intelligence and Data Science, University of Science and Technology of China, Hefei, China · Xiaohongshu Inc., Beijing, China · Thrust of Artificial Intelligence, The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, China +1
Efficient, Property-Aligned Fan-Out Retrieval via RL-Compiled Diffusion
Mar 6, 2026Pengcheng Jiang, Judith Yue Li, Moonkyung Ryu +8Information Retrieval Generative Retrieval
Google Research · University of Illinois Urbana Champaign
Vectorizing the Trie: Efficient Constrained Decoding for LLM-based Generative Retrieval on Accelerators
Feb 26, 2026Zhengyang Su, Isay Katsman, Yueqi Wang +10Constrained Decoding Generative Retrieval
Google Mountain View, CA, USA · Yale University New Haven, CT, USA
Beyond Pipelines: A Fundamental Study on the Rise of Generative-Retrieval Architectures in Web Research
Feb 19, 2026Amirereza Abbasi, Mohsen HooshmandInformation Retrieval Generative Retrieval
Department of Computer Science and Information Technology Institute for Advanced Studies in Basic Sciences Zanjan, Iran

May 12, 2026J/K move · Enter open · S save

ORBIT: Preserving Foundational Language Capabilities in GenRetrieval via Origin-Regulated Merging

Neha Verma, Nikhil Mehta, Shao-Chuan Wang, Naijing Zhang+6

Johns Hopkins University · Google DeepMind · Google

Continual Learning for LLMs LLM Fine-Tuning Generative Retrieval Catastrophic Forgetting Fine-Tuning Language Models

Open paper AskPDF arXiv

Despite the rapid advancements in large language model (LLM) development, fine-tuning them for specific tasks often results in the catastrophic forgetting of their general, language-based reasoning abilities. This work investigates and addresses this challenge in the context of the Generative Retrieval (GenRetrieval) task. During GenRetrieval fine-tuning, we find this forgetting occurs rapidly and correlates with the distance between the fine-tuned and original model parameters. Given these observations, we propose ORBIT, a novel approach that actively tracks the distance between fine-tuned and initial model weights, and uses a weight averaging strategy to constrain model drift during GenRetrieval fine-tuning when this inter-model distance exceeds a maximum threshold. Our results show that ORBIT retains substantial text and retrieval performance by outperforming both common continual learning baselines and related regularization methods that also employ weight averaging.