Spend Search Where It Pays: Value-Guided Structured Sampling and Optimization for Generative Recommendation

TL;DR

V-STAR introduces a value-guided sampling framework for autoregressive generative recommendation, enhancing exploration and learning efficiency by addressing probability-reward mismatches and bias issues in decoding.

Contribution

It presents V-STAR, a novel framework combining value-guided decoding and sibling-relative advantage computation to improve generative recommendation performance.

Findings

Outperforms state-of-the-art baselines in accuracy and diversity.

Enhances exploration efficiency without exhaustive search.

Demonstrates effectiveness on offline and online datasets.

Abstract

Generative recommendation via autoregressive models has unified retrieval and ranking into a single conditional generation framework. However, fine-tuning these models with Reinforcement Learning (RL) often suffers from a fundamental probability-reward mismatch. Conventional likelihood-dominated decoding (e.g., beam search) exhibits a myopic bias toward locally probable prefixes, which causes two critical failures: (1) insufficient exploration, where high-reward items in low-probability branches are prematurely pruned and rarely sampled, and (2) advantage compression, where trajectories sharing high-probability prefixes receive highly correlated rewards with low within-group variance, yielding a weak comparative signal for RL. To address these challenges, we propose V-STAR, a Value-guided Sampling and Tree-structured Advantage Reinforcement framework. V-STAR forms a self-evolving loop via two synergistic components. First, a Value-Guided Efficient Decoding (VED) is developed to identify decisive nodes and selectively deepen high-potential prefixes. This improves exploration efficiency without exhaustive tree search. Second, we propose Sibling-GRPO, which exploits the induced tree topology to compute sibling-relative advantages and concentrates learning signals on decisive branching decisions. Extensive experiments on both offline and online datasets demonstrate that V-STAR outperforms state-of-the-art baselines, delivering superior accuracy and candidate-set diversity under strict latency constraints.