MixFormer: Co-Scaling Up Dense and Sequence in Industrial Recommenders

As industrial recommender systems enter a scaling-driven regime, Transformer architectures have become increasingly attractive for scaling models towards larger capacity and longer sequence. However, existing Transformer-based recommendation models remain structurally fragmented, where sequence modeling and feature interaction are implemented as separate modules with independent parameterization. Such designs introduce a fundamental co-scaling challenge, as model capacity must be suboptimally allocated between dense feature interaction and sequence modeling under a limited computational budget. In this work, we propose MixFormer, a unified Transformer-style architecture tailored for recommender systems, which jointly models sequential behaviors and feature interactions within a single backbone. Through a unified parameterization, MixFormer enables effective co-scaling across both dense capacity and sequence length, mitigating the trade-off observed in decoupled designs. Moreover, the integrated architecture facilitates deep interaction between sequential and non-sequential representations, allowing high-order feature semantics to directly inform sequence aggregation and enhancing overall expressiveness. To ensure industrial practicality, we further introduce a user-item decoupling strategy for efficiency optimizations that significantly reduce redundant computation and inference latency. Extensive experiments on large-scale industrial datasets demonstrate that MixFormer consistently exhibits superior accuracy and efficiency. Furthermore, large-scale online A/B tests on two production recommender systems, Douyin and Douyin Lite, show consistent improvements in user engagement metrics, including active days and in-app usage duration.

DMGIN: How Multimodal LLMs Enhance Large Recommendation Models for Lifelong User Post-click Behaviors

Modeling user interest based on lifelong user behavior sequences is crucial for enhancing Click-Through Rate (CTR) prediction. However, long post-click behavior sequences themselves pose severe performance issues: the sheer volume of data leads to high computational costs and inefficiencies in model training and inference. Traditional methods address this by introducing two-stage approaches, but this compromises model effectiveness due to incomplete utilization of the full sequence context. More importantly, integrating multimodal embeddings into existing large recommendation models (LRM) presents significant challenges: These embeddings often exacerbate computational burdens and mismatch with LRM architectures. To address these issues and enhance the model's efficiency and accuracy, we introduce Deep Multimodal Group Interest Network (DMGIN). Given the observation that user post-click behavior sequences contain a large number of repeated items with varying behaviors and timestamps, DMGIN employs Multimodal LLMs(MLLM) for grouping to reorganize complete lifelong post-click behavior sequences more effectively, with almost no additional computational overhead, as opposed to directly introducing multimodal embeddings. To mitigate the potential information loss from grouping, we have implemented two key strategies. First, we analyze behaviors within each group using both interest statistics and intra-group transformers to capture group traits. Second, apply inter-group transformers to temporally ordered groups to capture the evolution of user group interests. Our extensive experiments on both industrial and public datasets confirm the effectiveness and efficiency of DMGIN. The A/B test in our LBS advertising system shows that DMGIN improves CTR by 4.7% and Revenue per Mile by 2.3%.

Deep Multiple Quantization Network on Long Behavior Sequence for Click-Through Rate Prediction

In Click-Through Rate (CTR) prediction, the long behavior sequence, comprising the user's long period of historical interactions with items has a vital influence on assessing the user's interest in the candidate item. Existing approaches strike efficiency and effectiveness through a two-stage paradigm: first retrieving hundreds of candidate-related items and then extracting interest intensity vector through target attention. However, we argue that the discrepancy in target attention's relevance distribution between the retrieved items and the full long behavior sequence inevitably leads to a performance decline. To alleviate the discrepancy, we propose the Deep Multiple Quantization Network (DMQN) to process long behavior sequence end-to-end through compressing the long behavior sequence. Firstly, the entire spectrum of long behavior sequence will be quantized into multiple codeword sequences based on multiple independent codebooks. Hierarchical Sequential Transduction Unit is incorporated to facilitate the interaction of reduced codeword sequences. Then, attention between the candidate and multiple codeword sequences will output the interest vector. To enable online serving, intermediate representations of the codeword sequences are cached, significantly reducing latency. Our extensive experiments on both industrial and public datasets confirm the effectiveness and efficiency of DMQN. The A/B test in our advertising system shows that DMQN improves CTR by 3.5% and RPM by 2.0%.