PROMISE: Process Reward Models Unlock Test-Time Scaling Laws in Generative Recommendations
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_ | 1866915715877961728 |
|---|---|
| author | Guo, Chengcheng Cai, Kuo Zhou, Yu Luo, Qiang Tang, Ruiming Li, Han Gai, Kun Zhou, Guorui |
| author_facet | Guo, Chengcheng Cai, Kuo Zhou, Yu Luo, Qiang Tang, Ruiming Li, Han Gai, Kun Zhou, Guorui |
| contents | Generative Recommendation has emerged as a promising paradigm, reformulating recommendation as a sequence-to-sequence generation task over hierarchical Semantic IDs. However, existing methods suffer from a critical issue we term Semantic Drift, where errors in early, high-level tokens irreversibly divert the generation trajectory into irrelevant semantic subspaces. Inspired by Process Reward Models (PRMs) that enhance reasoning in Large Language Models, we propose Promise, a novel framework that integrates dense, step-by-step verification into generative models. Promise features a lightweight PRM to assess the quality of intermediate inference steps, coupled with a PRM-guided Beam Search strategy that leverages dense feedback to dynamically prune erroneous branches. Crucially, our approach unlocks Test-Time Scaling Laws for recommender systems: by increasing inference compute, smaller models can match or surpass larger models. Extensive offline experiments and online A/B tests on a large-scale platform demonstrate that Promise effectively mitigates Semantic Drift, significantly improving recommendation accuracy while enabling efficient deployment. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2601_04674 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | PROMISE: Process Reward Models Unlock Test-Time Scaling Laws in Generative Recommendations Guo, Chengcheng Cai, Kuo Zhou, Yu Luo, Qiang Tang, Ruiming Li, Han Gai, Kun Zhou, Guorui Information Retrieval Generative Recommendation has emerged as a promising paradigm, reformulating recommendation as a sequence-to-sequence generation task over hierarchical Semantic IDs. However, existing methods suffer from a critical issue we term Semantic Drift, where errors in early, high-level tokens irreversibly divert the generation trajectory into irrelevant semantic subspaces. Inspired by Process Reward Models (PRMs) that enhance reasoning in Large Language Models, we propose Promise, a novel framework that integrates dense, step-by-step verification into generative models. Promise features a lightweight PRM to assess the quality of intermediate inference steps, coupled with a PRM-guided Beam Search strategy that leverages dense feedback to dynamically prune erroneous branches. Crucially, our approach unlocks Test-Time Scaling Laws for recommender systems: by increasing inference compute, smaller models can match or surpass larger models. Extensive offline experiments and online A/B tests on a large-scale platform demonstrate that Promise effectively mitigates Semantic Drift, significantly improving recommendation accuracy while enabling efficient deployment. |
| title | PROMISE: Process Reward Models Unlock Test-Time Scaling Laws in Generative Recommendations |
| topic | Information Retrieval |
| url | https://arxiv.org/abs/2601.04674 |