Climber-Pilot: A Non-Myopic Generative Recommendation Model Towards Better Instruction-Following

2026-02-14

AI Summary

This work addresses the short-sightedness of existing generative recommendation models, which struggle to capture users’ long-term, multi-item consumption intents and efficiently adhere to business-driven constraints such as category control. To overcome these limitations, the authors propose a unified generative retrieval framework that introduces a novel training paradigm—Time-Aware Multi-Item Prediction (TAMIP)—to model long-horizon user intent. Additionally, they design a Condition-Guided Sparse Attention (CGSA) mechanism that directly incorporates business constraints into the generation process, enabling precise instruction following without incurring any additional inference overhead. Deployed on NetEase Cloud Music, the proposed method achieves a 4.24% improvement in core business metrics, significantly outperforming state-of-the-art baselines.

Technology Category

• Natural Language Processing: Generation
• Machine Learning: Deep Generative Models & Autoencoders
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications

Abstract

Generative retrieval has emerged as a promising paradigm in recommender systems, offering superior sequence modeling capabilities over traditional dual-tower architectures. However, in large-scale industrial scenarios, such models often suffer from inherent myopia: due to single-step inference and strict latency constraints, they tend to collapse diverse user intents into locally optimal predictions, failing to capture long-horizon and multi-item consumption patterns. Moreover, real-world retrieval systems must follow explicit retrieval instructions, such as category-level control and policy constraints. Incorporating such instruction-following behavior into generative retrieval remains challenging, as existing conditioning or post-hoc filtering approaches often compromise relevance or efficiency. In this work, we present Climber-Pilot, a unified generative retrieval framework to address both limitations. First, we introduce Time-Aware Multi-Item Prediction (TAMIP), a novel training paradigm designed to mitigate inherent myopia in generative retrieval. By distilling long-horizon, multi-item foresight into model parameters through time-aware masking, TAMIP alleviates locally optimal predictions while preserving efficient single-step inference. Second, to support flexible instruction-following retrieval, we propose Condition-Guided Sparse Attention (CGSA), which incorporates business constraints directly into the generative process via sparse attention, without introducing additional inference steps. Extensive offline experiments and online A/B testing at NetEase Cloud Music, one of the largest music streaming platforms, demonstrate that Climber-Pilot significantly outperforms state-of-the-art baselines, achieving a 4.24\% lift of the core business metric.

Problem

• generative retrieval
• myopia
• instruction-following
• multi-item prediction
• retrieval constraints

Innovation

• generative retrieval
• non-myopic recommendation
• instruction-following
• time-aware masking
• sparse attention