TransRetrieval：面向工业推荐的基于Transformer的检索框架的规模化
TransRetrieval: Scaling Up Transformer-Based Retrieval for Industrial Recommendation
浏览论文内容
中文总结 AI 辅助
针对工业推荐检索的特征异质性问题，提出TransRetrieval框架，通过加权平均聚合、目标token压缩等技术实现高效缩放，在数据集和基准测试中取得召回率提升，线上测试也实现收益增长。
中文摘要 AI 辅助
将缩放定律应用于推荐检索受到特征异质性的阻碍：直接堆叠Transformer层会产生收益递减，因为异构字段会导致严重的token范数发散。我们提出了TransRetrieval，一种基于Transformer的检索框架，可随计算预算和跨域数据进行缩放。关键的实现手段是：（1）加权平均聚合，恢复Transformer所依赖的同构token假设；在此基础上，我们引入（2）目标token压缩，将每个候选样本的浮点运算量（FLOPs）降低85%，同时保留交叉注意力的表达能力；以及（3）位置式域嵌入，以可忽略的额外成本统一多个域，将跨域数据转化为可缩放的资产。在包含400亿次交互的工业数据集和公开的KuaiRand基准测试中，当每个目标的计算量从0.1 MFLOPs提升至2 MFLOPs时，Recall@2000分别提升19.3和22.2个百分点，证实了稳健的对数线性缩放特性。在线A/B测试中，在与生产基线相同的端到端延迟约束下，TransRetrieval使平台收益提升2.53%。
英文摘要
Applying scaling laws to recommendation retrieval is hindered by feature heterogeneity: naively stacking Transformer layers yields diminishing returns because heterogeneous fields produce severe token-norm divergence. We present TransRetrieval, a Transformer-based retrieval framework that scales with both computational budget and cross-domain data. The key enabler is (1) weighted average aggregation, which restores the homogeneous-token assumption Transformers rely on. Building on this, we introduce (2) target token compression that cuts per-candidate FLOPs by 85% while preserving cross-attention expressiveness, and (3) position-style domain embeddings that unify multiple domains at negligible additional cost, turning cross-domain data into a scaling asset. On a 40-billion-interaction industrial dataset and the public KuaiRand benchmark, scaling compute from 0.1 to 2 MFLOPs per target yields +19.3/+22.2 pt Recall@2000, confirming robust log-linear scaling. In online A/B tests, TransRetrieval lifts platform revenue by 2.53% under the same end-to-end latency constraint as the production baseline.
