LaRec: Unleashing LLM-based Latent Reasoning for Generative Recommendation

2026-07-27

AI Summary

This work addresses the limitations of existing large language model–based recommendation approaches, which either rely on verbose explicit chains of thought or are constrained by sparse supervision and single-path assumptions in implicit latent-space reasoning. To overcome these issues, we propose LaRec, a framework that enables efficient multi-path latent reasoning through a two-stage training paradigm. First, the model is pretrained in the latent space; then, fine-grained supervision is introduced via step-level and process-direction alignment. Additionally, a personalized Gaussian mixture distribution is constructed for each user to capture diverse interests. Finally, reinforcement learning–based fine-tuning further optimizes recommendation performance. Experiments demonstrate that LaRec significantly outperforms state-of-the-art methods across multiple benchmark datasets while maintaining computational efficiency during inference.

Abstract

Large Language Models (LLMs) have shown great promise in recommendation due to superior reasoning abilities. However, existing methods mainly rely on explicit Chain-of-Thought (CoT), resulting in verbose reasoning texts and inefficient response times. latent reasoning aims to balance efficiency by thinking within a continuous latent space, yet it faces two major challenges: (1) Lack of Fine-grained Supervision: Latent reasoning relies solely on feedback from the final labels, providing sparse supervisory signals that struggle to effectively guide the optimization of multiple hidden reasoning steps. (2) Single Reasoning Path: The deterministic nature of latent reasoning impedes the exploration of users' diverse interests and preferences, thereby limiting the recommendation capabilities of LLMs. To address these issues, we propose \textbf{$LaRec$}, an efficient generative recommendation framework designed to unleash the potential of latent reasoning in LLMs. $LaRec$ consists of two core stages: First, we design Latent Pre-training that empowers LLMs with latent reasoning capabilities by providing rich supervisory signals to the latent space reasoning via step-level alignment and process direction alignment. Second, we introduce Personalized RL-tuning. Specifically, we construct a personalized Gaussian Mixture Distribution for each user based on their historical interests. By randomly sampling distinct reasoning starting points from this distribution during training, we guide the LLMs to traverse diverse reasoning paths within the latent space, enabling efficient exploration of user's multi-faceted interests. Experiments on multiple datasets show that $LaRec$ significantly outperforms existing baselines with comparable efficiency.

Problem

• latent reasoning
• fine-grained supervision
• single reasoning path
• generative recommendation
• Large Language Models

Innovation

• Latent Reasoning
• Generative Recommendation
• Fine-grained Supervision
• Personalized RL-tuning
• Gaussian Mixture Distribution