Search-to-list distillation improves single-pass generative item retrieval
SLIDE: Search-to-List Item Distillation for Single-Rollout Generative Retrieval
Information Retrieval
Summary
Finding good items from a large collection is important but challenging. The authors noticed that generating a list of items one by one usually performs worse than searching multiple possibilities separately and then combining results. They created a method called SLIDE that learns from a strong search process to generate a full list in one pass, improving retrieval quality. This approach lets systems find better item lists faster without needing complex search during use.
What this means in practice
- •For ecommerce platform engineers: Enhance product search by generating whole relevant product lists in one step, improving speed and quality without complex online searches.$Commercial implications: Allows ecommerce services to deliver faster and more accurate product recommendations, improving user experience and conversion rates.
- •For digital advertising teams: Improve ad retrieval systems to generate targeted ad lists efficiently, enabling better personalization while reducing runtime computation.
Authors
Tianyu Zhan, Gui Ling, Ran Tao, Kunhai Lin, Tong Xiong, Wang Yang, Kaixuan Zhang, Zhihong Chen, Yuliang Yan, Dan Ou, Shengyu Zhang, Haihong Tang
Abstract
Generative retrieval (GR) typically relies on item-level beam search to retrieve candidates, leaving cross-item dependencies unmodeled and limiting direct optimization of list-level objectives. A natural alternative is to generate the entire list within a single autoregressive trajectory. However, greedy decoding with directly trained list-wise models substantially underperforms beam search. We identify two contributing factors: the broader search space explored by beam search and the positional bias of scarce positive signals, which deprives later positions of adequate supervision. To address these challenges, we propose $\textbf{SLIDE}$, a search-to-list item distillation framework that decouples offline search from online list generation. A teacher model retrieves high-quality candidates through offline item-level beam search, providing additional conversion-relevant supervision beyond sparse behavioral labels. Their Semantic IDs (SIDs) are then concatenated in beam-score order to form list-wise distillation targets. This search-to-list conversion transforms candidates retrieved along separate search paths into sequential item-generation targets within a single trajectory. Trained on these targets, the student generates the Top-$K$ list in a single greedy pass, also decoupling teacher scaling from online deployment. Experiments on FORGE and Taobao Search show that SLIDE substantially improves retrieval quality under single greedy decoding and, with stronger offline teachers, surpasses the beam-search reference. Supplementary RL experiments support the effectiveness of list-level generation.