Search papers, labs, and topics across Lattice.
This paper introduces TmallGS, a novel scalable ranking architecture designed for generative e-commerce search that addresses the limitations of traditional DLRMs by leveraging advanced Transformer techniques. By incorporating components like Hierarchical Distribution-Calibrated Tokenization and a Field-Adaptive Gated Transformer Backbone, TmallGS enhances Model FLOPs Utilization and improves user engagement metrics such as UCTCVR and GMV. Extensive offline experiments and online A/B tests validate that TmallGS significantly boosts training throughput and ranking performance compared to existing models.
TmallGS achieves remarkable improvements in e-commerce search performance by effectively integrating diverse feature representations into a unified Transformer framework.
In industrial search and ranking systems, Click-Through Rate (CTR) prediction is shifting from traditional Deep Learning Recommendation Models (DLRM) toward unified, compute-intensive Transformer architectures. This transition is driven by the need to improve Model FLOPs Utilization (MFU) and achieve predictable gains through scaling laws. However, existing approaches such as OneTrans and Climber often adopt an all-in-tokenization strategy when adapting Large Language Model (LLM) architectures, overlooking the heterogeneous nature of ranking features. We propose TmallGS, a scalable ranking architecture for Tmall search. TmallGS includes five key components: (1) Hierarchical Distribution-Calibrated Tokenization, which combines Field-wise Saliency Reweighting (FSR) and Distribution-Calibrated Projection (DCP) to map diverse features into optimized subspaces; (2) a Field-Adaptive Gated Transformer Backbone with per-field QKV projections and noise-adaptive gating for refined semantic interaction; (3) Decoupled FiLM Late Fusion to preserve explicit high-frequency signals; (4) a Context-Aware Bias Net to decouple systemic bias from user intent; and (5) Error-Aware Progressive Training with dynamically weighted losses for robust learning. Extensive offline experiments and online A/B tests on Tmall Search show that TmallGS improves training throughput and achieves substantial gains in UCTCVR and GMV.
DAMO