SimGR: Escaping the Pitfalls of Generative Decoding in LLM-based Recommendation
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866912888194596864 |
|---|---|
| author | Zhao, Yuanbo Liu, Ruochen Wang, Senzhang Yin, Jun Dong, Yuxin Gong, Huan Chen, Hao Pan, Shirui Zhang, Chengqi |
| author_facet | Zhao, Yuanbo Liu, Ruochen Wang, Senzhang Yin, Jun Dong, Yuxin Gong, Huan Chen, Hao Pan, Shirui Zhang, Chengqi |
| contents | A core objective in recommender systems is to accurately model the distribution of user preferences over items to enable personalized recommendations. Recently, driven by the strong generative capabilities of large language models (LLMs), LLM-based generative recommendation has become increasingly popular. However, we observe that existing methods inevitably introduce systematic bias when estimating item-level preference distributions. Specifically, autoregressive generation suffers from incomplete coverage due to beam search pruning, while parallel generation distorts probabilities by assuming token independence. We attribute this issue to a fundamental modeling mismatch: these methods approximate item-level distributions via token-level generation, which inherently induces approximation errors. Through both theoretical analysis and empirical validation, we demonstrate that token-level generation cannot faithfully substitute item-level generation, leading to biased item distributions. To address this, we propose \textbf{Sim}ply \textbf{G}enerative \textbf{R}ecommendation (\textbf{SimGR}), a framework that directly models item-level preference distributions in a shared latent space and ranks items by similarity, thereby aligning the modeling objective with recommendation and mitigating distributional distortion. Extensive experiments across multiple datasets and LLM backbones show that SimGR consistently outperforms existing generative recommenders. Our code is available at https://anonymous.4open.science/r/SimGR-C408/ |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_07847 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | SimGR: Escaping the Pitfalls of Generative Decoding in LLM-based Recommendation Zhao, Yuanbo Liu, Ruochen Wang, Senzhang Yin, Jun Dong, Yuxin Gong, Huan Chen, Hao Pan, Shirui Zhang, Chengqi Information Retrieval A core objective in recommender systems is to accurately model the distribution of user preferences over items to enable personalized recommendations. Recently, driven by the strong generative capabilities of large language models (LLMs), LLM-based generative recommendation has become increasingly popular. However, we observe that existing methods inevitably introduce systematic bias when estimating item-level preference distributions. Specifically, autoregressive generation suffers from incomplete coverage due to beam search pruning, while parallel generation distorts probabilities by assuming token independence. We attribute this issue to a fundamental modeling mismatch: these methods approximate item-level distributions via token-level generation, which inherently induces approximation errors. Through both theoretical analysis and empirical validation, we demonstrate that token-level generation cannot faithfully substitute item-level generation, leading to biased item distributions. To address this, we propose \textbf{Sim}ply \textbf{G}enerative \textbf{R}ecommendation (\textbf{SimGR}), a framework that directly models item-level preference distributions in a shared latent space and ranks items by similarity, thereby aligning the modeling objective with recommendation and mitigating distributional distortion. Extensive experiments across multiple datasets and LLM backbones show that SimGR consistently outperforms existing generative recommenders. Our code is available at https://anonymous.4open.science/r/SimGR-C408/ |
| title | SimGR: Escaping the Pitfalls of Generative Decoding in LLM-based Recommendation |
| topic | Information Retrieval |
| url | https://arxiv.org/abs/2602.07847 |