GR2 Technical Report

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: Yufei Li, Zaiwei Zhang, Mingfu Liang, Kavosh Asadi, Jay Xu, Jimmy Kim, Chongyang Bai, Jieyi Zhang, Hongye Xie, Prachi Agrawal, Dian Yu, Tianyi Chen, Jean-Pascal Billaud, Garret Buell, YK, Zhu, Sachin Patil, Brooke Bian, Zhou Fang, Kevin Huang, Shiva Sudanagunta, Yuzhen Huang, Emma Lu, Chris O'Brien, Yang Song, Lihong Li, Jacob Tao, Zhicheng Zhu, Chao Li, Gaoxiang Liu, Neil Wu, Zhongyin Hu, Li Han, Loki Chen, Ming Lei, Greg Rehm, Siyuan Song, Tianwei Zhang, Li Li, Ketan Singh, Yavuz Yetim, Ilyas Atishev, Satendra Gera, Ashkan Sadeghi, Rachel Yan, Nikko Mizutani, Shuaiwen Wang, Song Yang, Zhijing Li, Jiang Liu, Mengying Sun, Fei Tian, Xiaohan Wei, Chonglin Sun, Parish Aggarwal, Kaushik Rangadurai, Zhi Hua, Frank Shyu, Ruchit Sharma, Liyuan Li, Shike Mei, Wenlin Chen, Santanu Kolay, Ben Schulte, Deepak Chandra, Adam, Song, Sandeep Pandey, Xi Liu, Hamed Firooz, Luke Simon
• Venue: arXiv: cs.IR
• Published: 2026-06-30
• Collected: 2026-07-01
• URL: http://arxiv.org/abs/2606.31984v1
• PDF: https://arxiv.org/pdf/2606.31984v1
• DOI: UNKNOWN

Plain-Language Brief

Researchers present GR2 (Generative Reasoning Re-Ranker), a framework designed to improve the performance of recommendation systems in an industrial setting. The system combines mid-training on semantic IDs, reasoning-trace distillation from a stronger teacher, and reinforcement learning with verifiable rewards. It also includes a context compressor, On-Policy Distillation, and reasoning distillation for low-latency serving. GR2 outperforms legacy baselines in industrial-scale traffic by 18.7% R@1, 7.1% R@3, and 9.6% N@3.

What the Document Says

• GR2 is an end-to-end framework for improving recommendation systems.
• It combines mid-training on semantic IDs, reasoning-trace distillation, and reinforcement learning with verifiable rewards.
• Additional components include a context compressor, On-Policy Distillation, and reasoning distillation for low-latency serving.
• The researchers found that GR2 outperforms legacy baselines in industrial-scale traffic by 18.7% R@1, 7.1% R@3, and 9.6% N@3.

Key Concepts Mentioned

• recommendation systems
• industrial adoption
• Large Language Models (LLMs)
• reinforcement learning (RL)
• verifiable rewards
• GR2

Claims Made by the Document

• Most efforts in LLMs target retrieval and ranking, leaving re-ranking underexplored.
• Deployed catalogs have non-semantic identifiers outside any base-LLM vocabulary.
• GR2 delivers better performance than legacy baselines.

Evidence Used

• The text mentions that most efforts in LLMs target retrieval and ranking but leave re-ranking underexplored.
• The text states that deployed catalogs have non-semantic identifiers outside any base-LLM vocabulary.
• The text claims that GR2 delivers better performance than legacy baselines.

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Industrial recommendation systems serve billions of users through a multi-stage funnel -- retrieval, early-stage ranking, and re-ranking -- where the final re-ranking step disproportionately shapes user engagement and downstream performance, particularly for carousel and grid display formats. Despite growing enthusiasm for Large Language Models (LLMs) in recommendation, three gaps hinder industrial adoption: (1) most efforts target retrieval and ranking, leaving re-ranking -- the stage closest to the final user experience -- largely underexplored; (2) LLMs are typically deployed zero-shot or via supervised fine-tuning, underutilizing the reasoning capabilities unlocked by reinforcement learning (RL) on verifiable rewards; (3) deployed catalogs index billions of items with non-semantic identifiers that lie outside any base-LLM vocabulary. We present GR2 (Generative Reasoning Re-Ranker), an end-to-end framework that combines (i) mid-training on semantic IDs produced by a tokenizer with >=99% uniqueness, (ii) reasoning-trace distilled from a stronger teacher via targeted prompting and rejection sampling, and (iii) RL with verifiable rewards purpose-built for re-ranking. To make GR2 resource-viable, we further (iv) introduce a context compressor that amortizes training cost, On-Policy Distillation (OPD) as a scalable alternative to SFT -- which we find collapses at industrial scale -- and reasoning distillation for low-latency serving. GR2 delivers +18.7% R@1, +7.1% R@3, and +9.6% N@3 over legacy baselines on industrial-scale traffic. We further find that reward design is critical in re-ranking: LLMs often hack rewards by preserving the incoming order or exploiting position bias, motivating conditional verifiable rewards as essential industrial components.