Pith. sign in

REVIEW 3 major objections 6 minor 42 references

TransX reformulates recommendation as sequence-to-sequence action transduction over two separated streams — user behavior and serving events — and shows in production that the split, plus nearline caching, lifts click-through rate by 6% and

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · deepseek-v4-flash

2026-08-03 16:49 UTC pith:JGT6TYMW

load-bearing objection A credible industrial co-design paper with a clean complexity story and strong deployed results; the one real hole is that the cache-freshness assumption is asserted, not measured. the 3 major comments →

arxiv 2607.28940 v1 pith:JGT6TYMW submitted 2026-07-31 cs.IR

TransX: Scaling Transformer-based Recommendation via Behavioral and Serving Stream Crossings

Da Xu , Liyan Fang , Divya Venugopalan , Sunny Hsu , Xukai Wang , Rishav Roy Chowdhury , Cindy Liang , Nishant Satya Lakshmikanth This is my paper
classification cs.IR
keywords industrial recommender systemssequence-to-sequence transductionencoder-decoder transformercross-attentionstream crossingnearline amortized encodingKV cachingmodel-infrastructure co-design
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper argues that industrial recommenders should stop flattening all user signals into one token stream. It proposes TransX, an encoder–decoder model that keeps long-term user behavior and the system's immediate serving slate as separate streams and predicts each user action by 'crossing' the two with attention. The core claim is that this split is not only more faithful to the data's causal structure — user behavior evolves independently of system exposures — but is the key to making Transformer-based recommendation affordable: behavior encodings are computed nearline and cached, so online latency no longer depends on history length. On a large social recommendation platform, the paper reports a 6.0% click-through lift and 4.4% conversion gain over the production deep model, with online computation cut by roughly 80% and hardware cost comparable to the legacy model.

Core claim

TransX's central claim is that recommendation is best modeled as sequence-to-sequence action transduction: decode the joint action tokens for all candidates in a serving event, conditioned on the crossing of the user's behavior stream (encoded by a causal Transformer) and the serving stream (candidate features), using grouped multi-query sparse cross-attention with a local–global window. Because the behavior encoder uses causal attention, its outputs are prefix-consistent and can be computed incrementally nearline and cached; online, only the lightweight cross-attention and parallel action decoding run. This makes serving latency independent of behavioral history length, enabling a 180-day b

What carries the argument

The central mechanism is the 'stream crossing': a grouped multi-query sparse cross-attention where all candidates in a serving event share key/value representations from the cached behavior stream but keep candidate-specific queries, restricted to the most recent local tokens plus one global attention-pooled anchor. Around it sit three co-designed pieces: causal prefix reuse in the behavior encoder (one left-to-right pass, no re-encoding), nearline amortized encoding with per-user KV caching, and a parallel non-autoregressive decoder with a query–key softmax over a small action vocabulary. Together they turn an O(T·L²·d) training cost into O(L²·d + T) and remove L from online complexity.

Load-bearing premise

The reported quality and latency gains rest on the premise that nearline-computed, cached behavior encodings remain accurate enough between refreshes; if stale encodings degrade prediction quality, the latency savings trade away the CTR/CVR lifts.

What would settle it

In the same production A/B setup, artificially freeze or delay the nearline refresh of cached behavior encodings for an extended period (e.g., skip all updates for 24 hours) and measure the CTR and conversion deltas. If the +6.0% CTR lift collapses while the ~80% latency reduction stays, the freshness assumption is the load-bearing wall.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X Bluesky LinkedIn Reddit HN

If this is right

  • Long user histories (e.g., 180 days) become practical in production ranking without a proportional rise in serving cost.
  • Other sequential and generative models that re-encode behavior per event become comparatively more expensive, shifting the design target toward amortized encoding.
  • The same cached behavior encodings can be reused across multiple downstream tasks, since they are learned independent of any serving event.
  • The architecture suggests a path to a unified retrieval-and-ranking model by adding a contrastive objective between behavior and candidate encodings (the paper notes this itself).
  • Model–infrastructure co-design, rather than a model-only change, is what lets a Transformer ranker match the serving economics of a DLRM.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The decisive test the paper leaves open is cache freshness: no experiment systematically ages the cached behavior encodings. Measuring CTR lift against refresh interval would show whether the +6% is robust or partly an artifact of near-real-time updates.
  • The stream-crossing principle may generalize to other two-sided systems (advertising, news, jobs feeds) where user intent and system exposure are causally distinct; a transfer study would test that.
  • Because behavior encodings are cached independently of the action head, the same encoder could serve multiple recommendation tasks (e.g., CTR and CVR) with one shared nearline cost, a multiplier the paper does not quantify.
  • The O(L²·d + T) training complexity claim implies that as T grows, training cost stays roughly flat for a fixed user; verifying that on a larger corpus would confirm the amortization claim beyond the reported ~50% epoch-speedup.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper proposes TransX, an encoder–decoder recommendation architecture that decouples the user behavior stream from the serving-event stream, using sparse cross-attention between cached behavior encodings and per-request candidate queries, followed by a parallel non-autoregressive action decoder. The authors claim that this design enables long (180-day) behavior windows at production scale, with offline and online experiments on LinkedIn data showing consistent gains over DLRM and sequential/generative baselines, a +6.0% CTR lift and +4.4% conversion lift in A/B tests, and an ~80% reduction in online serving computation via amortized nearline encoding and KV caching. The appendix provides complexity analyses, deployment details, and additional latency and segment results.

Significance. If the empirical claims hold, TransX is a significant industrial contribution: it demonstrates a concrete architectural route to making long-horizon Transformer sequence modeling affordable for high-QPS recommendation, while also improving engagement and conversion. The paper's strengths include a clear separation of behavior and serving streams, a well-motivated sparse cross-attention design with shared key–value caching, a transparent complexity analysis (Appendix A) that correctly identifies the amortization of behavior encoding, and a fairly extensive ablation suite (Tables 2, 5, 6) supporting the main architectural choices. The reported online A/B results include statistical significance and segment analyses, which is more than many industrial papers provide. The main limitation is the lack of public data/code and the reliance on proprietary benchmarks, but that is not disqualifying for a KDD industrial track paper.

major comments (3)
  1. [Appendix E, 'Overall Takeaway'; Appendix D, 'Nearline update'] The central joint claim—that the ~80% online computation reduction is cost-neutral relative to the CTR/CVR gains—rests on the assumption that cached nearline behavior encodings remain sufficiently fresh. Appendix E asserts that gains 'hold under ... delayed cache refresh' but gives no refresh-lag distribution, no staleness measurement, and no experiment that intentionally varies cache freshness. This is load-bearing because Table 2 shows the behavior encoder is the dominant accuracy component (removing it degrades AUC by −28.3%), so stale encodings could plausibly erode the reported lifts. The paper should either report measured refresh latencies and their effect on online metrics, or run a controlled experiment that degrades freshness and shows the quality impact is bounded.
  2. [§5.3, Table 3; Appendix D, 'Nearline update'] The nearline pipeline is described as a mixture of a Kafka-based streaming path for 'highly active users' and an HDFS batch path for others, with no quantitative description of what fraction of users/traffic falls into each path, what the typical update latency is for the batch path, or how the 2% cold-start fallback interacts with the reported gains. Since the headline latency reduction is benchmarked at an average 80K requests per second under real traffic, the composition of refresh mechanisms directly affects whether the reported p99 latency and the accuracy gains hold for the full population. At minimum, the paper should report the traffic coverage of the streaming vs. batch nearline paths and the associated update intervals.
  3. [§4.4, Complexity Analysis; Appendix A] The training-complexity comparison is presented as reducing O(T L^2 d) to O(L^2 d + T), but Appendix A's more precise formula includes T m L_local d + T m k d terms. The §4.4 simplification is acceptable only if the event-level terms are negligible, which is not established. Given that m can be ~500 and L_local=10, T m L_local d may dominate L^2 d in practice. The authors should state the parameter regimes under which the claimed asymptotic reduction is meaningful, or present the fuller expression in the main text.
minor comments (6)
  1. [§5.1, Table 2] MFLOPs are reported in a compressed notation (e.g., '3.8/0.6×10^2') without units or a clear definition of what one 'candidate item scoring' includes. Clarify the exact operation count and whether it covers feature processing, attention, and decoding for a single candidate.
  2. [§5.3, Table 4] The claim that the CTR and conversion lifts are 'the largest observed in recent years' is not substantiated with any historical baseline or reference. Either provide the comparison context or soften the claim.
  3. [Appendix C] For the baselines, 'applying the same causal mask and local attention setup when applicable' is vague, especially for GRM and TransAct variants. Specify exactly which components were modified and whether any baseline was given an unfair disadvantage (e.g., forcing a local window on a model designed for full attention).
  4. [§5.2, Table 2 lower panel] Ablation metrics are reported as relative changes from TransX (e.g., −13.4% AUC), which is non-standard for AUC. Report absolute values or both; otherwise the magnitude of change is hard to interpret.
  5. [Appendix D, 'Remark: incremental encoding'] The claim that relative positional bias allows cached representations of historical tokens to remain unchanged under incremental updates is plausible but should be stated more carefully: while the representations of old tokens do not change, the global anchor and any position-dependent interactions may need renormalization. Clarify the exact update procedure.
  6. [Throughout] There are minor typographical issues (e.g., 'behaivor', 'benchmared', 'medium sequence length of the segments') that should be corrected in a camera-ready version.

Circularity Check

0 steps flagged

No significant circularity: TransX's claims rest on measured benchmarks and architectural complexity analysis, not on fitted inputs renamed as predictions.

full rationale

The paper does not derive any predicted result from a fitted constant or from a self-citation chain. Its central formulation (Eq. 4) is a conditional action-transduction objective, and the training loss (Eq. 13) is a standard event-factorized cross-entropy; neither defines the target quantity in terms of the quantity it is supposed to predict. Offline results (Table 2), ablations, and online A/B metrics are reported measurements on held-out test data and live traffic, so they are empirical claims rather than circular derivations. Complexity claims in Appendix A are closed-form consequences of the stated sparse-attention and caching design, not fitted results. Hyperparameter choices such as L_local=10 and two encoder layers are validation-set model selection, which is normal practice and not circular. The cited related work (LiRank, TransAct, GRM, etc.) consists of external baselines and prior systems; none of the load-bearing claims are justified only by a citation to the present authors' own prior work, and no uniqueness theorem or ansatz is imported from a self-citation. The only notable weakness is Appendix E's assertion that gains 'hold under ... delayed cache refresh' without reporting staleness measurements; that is missing evidence for a deployment claim, not a definitional equivalence or a fitted-parameter-renamed-as-prediction. Therefore no specific circular step can be exhibited, and the appropriate finding is no significant circularity with score 0.

Axiom & Free-Parameter Ledger

3 free parameters · 3 axioms · 0 invented entities

The paper's central result is a measured system-level performance, not a derivation from postulates. The only axioms are standard Transformer mathematics, the causal-mask prefix-reuse property, and operational freshness assumptions about the nearline cache. Free parameters are the usual architectural hyperparameters and the training downsampling ratio; none are fitted to construct a theoretical prediction.

free parameters (3)
  • Local cross-attention window L_local = 10
    Chosen by ablation over {5,10,15,20} in Table 6; trades compute against accuracy and is not derived from first principles.
  • Number of behavior encoder layers = 2
    Fixed to two Transformer blocks for capacity matching across all baselines; not derived from an optimality condition.
  • Training negative downsampling ratio = 1:1
    No-action tokens are downsampled to match positives during training (Section 5.1); this changes the training distribution and is a hand-chosen preprocessing choice.
axioms (3)
  • domain assumption Causal self-attention with relative position bias permits incremental nearline encoding without recomputing cached prefix representations.
    Appendix D 'Remark: incremental encoding' relies on this property. It is true for causal masking in isolation, but production correctness also depends on the implementation details of layer norm, residual connections, and relative position indexing.
  • domain assumption A 180-day behavior window captures the user intent relevant to the recommendation task.
    The dataset construction fixes 180-day lookback; the paper gives no analysis showing where the marginal value of longer history saturates.
  • domain assumption Nearline-cached behavior encodings remain fresh enough between event- and time-based refreshes to preserve the reported online gains.
    Appendix D's event-based and time-based cache updates, and the 'delayed cache refresh' robustness claim, assume staleness does not materially hurt the treatment effect; no staleness measurement is reported.

pith-pipeline@v1.3.0-daily-deepseek · 17374 in / 12853 out tokens · 111426 ms · 2026-08-03T16:49:32.762741+00:00 · methodology

0 comments
Cite this review

Pith. "Pith review of TransX: Scaling Transformer-based Recommendation via Behavioral and Serving Stream Crossings." pith.science (2026). https://pith.science/paper/JGT6TYMW

@misc{pith2026260728940,
  author       = {Pith},
  title        = {Pith review of: TransX: Scaling Transformer-based Recommendation via Behavioral and Serving Stream Crossings},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JGT6TYMW}},
  note         = {Machine review of arXiv:2607.28940}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Modern industrial recommender systems (RecSys) increasingly adopt Transformer-based sequence models, with an emerging paradigm that frames recommendation as next-token prediction over a unified monolithic user sequence. However, collapsing heterogeneous data sources -- such as long-term user behaviors and real-time serving events -- into a single monolithic token stream that obscures their distinct causal roles and temporal characteristics, leading to inefficient modeling and elevated training and serving costs. We propose TransX, a production-oriented encoder-decoder architecture that reformulates recommendation as a sequence-to-sequence action transduction problem. TransX explicitly decouples behavior-stream modeling from serving-event modeling and conditions next-action decoding on scalable cross-attention between nearline behavior encodings and real-time serving representations. To enable low-latency, high-QPS deployment, TransX is co-designed with an amortized serving strategy that combines incremental behavior encoding with per-request key-value caching, rendering serving latency insensitive to behavior sequence length. Extensive offline experiments and large-scale online A/B tests on LinkedIn's recommender systems show that TransX consistently outperforms state-of-the-art DLRMs and sequential baselines, and delivers substantial CTR lift (+6.0%) and conversion gain (+4.4%) while maintaining serving costs comparable to existing production models where our co-designed serving strategy reduces online computation by approximately 80%.

Figures

Figures reproduced from arXiv: 2607.28940 by Cindy Liang, Da Xu, Divya Venugopalan, Liyan Fang, Nishant Satya Lakshmikanth, Rishav Roy Chowdhury, Sunny Hsu, Xukai Wang.

Figure 1
Figure 1. Figure 1: End-to-end system overview of TransX, illustrating the unified design across offline training and online serving. Left: offline sequence-to-sequence training with a single forward–backward pass per user. Right: At online serving time, candidate queries attend to cached user behavior representations via grouped multi-query cross attention, followed by parallel (amortized) non-autoregressive action decoding.… view at source ↗
Figure 2
Figure 2. Figure 2: Overview of TransX’s online inference pipeline with model–infra co-design. User behavior streams are encoded of￾fline/nearline, cached per user, and incrementally updated near￾line. At serving time, only lightweight event-level computation is performed: candidate queries attend to a cached, local–global sub￾set of behavior representations via grouped multi-query cross at￾tention with shared key–value cache… view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

42 extracted references · 8 linked inside Pith

  1. [1]

    Joshua Ainslie, James Lee-Thorp, Michiel De Jong, Yury Zemlyanskiy, Federico Lebrón, and Sumit Sanghai. 2023. Gqa: Training generalized multi-query trans- former models from multi-head checkpoints.arXiv preprint arXiv:2305.13245 (2023)

    Pith/arXiv arXiv 2023

  2. [2]

    Fedor Borisyuk, Mingzhou Zhou, Qingquan Song, Siyu Zhu, Birjodh Tiwana, Ganesh Parameswaran, Siddharth Dangi, Lars Hertel, Qiang Charles Xiao, Xi- aochen Hou, et al . 2024. LiRank: Industrial Large Scale Ranking Models at LinkedIn. InProceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 4804–4815

    2024

  3. [3]

    Zheng Chai, Qin Ren, Xijun Xiao, Huizhi Yang, Bo Han, Sijun Zhang, Di Chen, Hui Lu, Wenlin Zhao, Lele Yu, et al . 2025. Longer: Scaling up long sequence modeling in industrial recommenders. InProceedings of the Nineteenth ACM Conference on Recommender Systems. 247–256

    2025

  4. [4]

    Qiwei Chen, Huan Zhao, Wei Li, Pipei Huang, and Wenwu Ou. 2019. Behavior sequence transformer for e-commerce recommendation in alibaba. InProceedings of the 1st international workshop on deep learning practice for high-dimensional sparse data. 1–4

    2019

  5. [5]

    Heng-Tze Cheng, Levent Koc, Jeremiah Harmsen, Tal Shaked, Tushar Chandra, Hrishi Aradhye, Glen Anderson, Greg Corrado, Wei Chai, Mustafa Ispir, et al

  6. [6]

    Zihang Dai, Zhilin Yang, Yiming Yang, Jaime G Carbonell, Quoc Le, and Ruslan Salakhutdinov. 2019. Transformer-xl: Attentive language models beyond a fixed- length context. InProceedings of the 57th annual meeting of the association for computational linguistics. 2978–2988

    2019

  7. [7]

    Tri Dao, Dan Fu, Stefano Ermon, Atri Rudra, and Christopher Ré. 2022. Flashat- tention: Fast and memory-efficient exact attention with io-awareness.Advances in neural information processing systems35 (2022), 16344–16359

    2022

  8. [8]

    Jiaxin Deng, Shiyao Wang, Kuo Cai, Lejian Ren, Qigen Hu, Weifeng Ding, Qiang Luo, and Guorui Zhou. 2025. Onerec: Unifying retrieve and rank with generative recommender and iterative preference alignment.arXiv preprint arXiv:2502.18965 (2025)

    Pith/arXiv arXiv 2025

  9. [9]

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. Bert: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 conference of the North American chapter of the association for computational linguistics: human language technologies, volume 1 (long and short papers). 4171–4186

    2019

  10. [10]

    Shijie Geng, Shuchang Liu, Zuohui Fu, Yingqiang Ge, and Yongfeng Zhang. 2022. Recommendation as language processing (rlp): A unified pretrain, personalized prompt & predict paradigm (p5). InProceedings of the 16th ACM conference on recommender systems. 299–315

    2022

  11. [11]

    Huan Gui, Ruoxi Wang, Ke Yin, Long Jin, Maciej Kula, Taibai Xu, Lichan Hong, and Ed H Chi. 2023. Hiformer: Heterogeneous feature interactions learning with transformers for recommender systems.arXiv preprint arXiv:2311.05884(2023)

    Pith/arXiv arXiv 2023

  12. [12]

    Huifeng Guo, Ruiming Tang, Yunming Ye, Zhenguo Li, and Xiuqiang He. 2017. DeepFM: a factorization-machine based neural network for CTR prediction.arXiv preprint arXiv:1703.04247(2017)

    Pith/arXiv arXiv 2017

  13. [13]

    Ruidong Han, Bin Yin, Shangyu Chen, He Jiang, Fei Jiang, Xiang Li, Chi Ma, Mincong Huang, Xiaoguang Li, Chunzhen Jing, et al . 2025. Mtgr: Industrial- scale generative recommendation framework in meituan. InProceedings of the 34th ACM International Conference on Information and Knowledge Management. 5731–5738

    2025

  14. [14]

    Xinran He, Junfeng Pan, Ou Jin, Tianbing Xu, Bo Liu, Tao Xu, Yanxin Shi, Antoine Atallah, Ralf Herbrich, Stuart Bowers, et al. 2014. Practical lessons from predicting clicks on ads at facebook. InProceedings of the eighth international workshop on data mining for online advertising. 1–9

    2014

  15. [15]

    Wang-Cheng Kang and Julian McAuley. 2018. Self-attentive sequential recom- mendation. In2018 IEEE international conference on data mining (ICDM). IEEE, 197–206

    2018

  16. [16]

    Lei Li, Yongfeng Zhang, Dugang Liu, and Li Chen. 2024. Large language models for generative recommendation: A survey and visionary discussions. InProceedings of the 2024 joint international conference on computational linguistics, language resources and evaluation (LREC-COLING 2024). 10146–10159

    2024

  17. [17]

    Jianxun Lian, Xiaohuan Zhou, Fuzheng Zhang, Zhongxia Chen, Xing Xie, and Guangzhong Sun. 2018. xdeepfm: Combining explicit and implicit feature in- teractions for recommender systems. InProceedings of the 24th ACM SIGKDD international conference on knowledge discovery & data mining. 1754–1763

    2018

  18. [18]

    Jiasen Lu, Dhruv Batra, Devi Parikh, and Stefan Lee. 2019. Vilbert: Pretrain- ing task-agnostic visiolinguistic representations for vision-and-language tasks. Advances in neural information processing systems32 (2019)

    2019

  19. [19]

    Maxim Naumov, Dheevatsa Mudigere, Hao-Jun Michael Shi, Jianyu Huang, Narayanan Sundaraman, Jongsoo Park, Xiaodong Wang, Udit Gupta, Carole- Jean Wu, Alisson G Azzolini, et al. 2019. Deep learning recommendation model for personalization and recommendation systems.arXiv preprint arXiv:1906.00091 (2019)

    Pith/arXiv arXiv 2019

  20. [20]

    Qi Pi, Guorui Zhou, Yujing Zhang, Zhe Wang, Lejian Ren, Ying Fan, Xiaoqiang Zhu, and Kun Gai. 2020. Search-based user interest modeling with lifelong sequential behavior data for click-through rate prediction. InProceedings of the 29th ACM International Conference on Information & Knowledge Management. 2685–2692

    2020

  21. [21]

    Jiarui Qin, Weinan Zhang, Xin Wu, Jiarui Jin, Yuchen Fang, and Yong Yu. 2020. User behavior retrieval for click-through rate prediction. InProceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval. 2347–2356

    2020

  22. [22]

    Shashank Rajput, Nikhil Mehta, Anima Singh, Raghunandan Hulikal Keshavan, Trung Vu, Lukasz Heldt, Lichan Hong, Yi Tay, Vinh Tran, Jonah Samost, et al

  23. [23]

    Francesco Ricci, Lior Rokach, and Bracha Shapira. 2021. Recommender systems: Techniques, applications, and challenges.Recommender systems handbook(2021), 1–35

    2021

  24. [24]

    Noam Shazeer. 2019. Fast transformer decoding: One write-head is all you need. arXiv preprint arXiv:1911.02150(2019)

    Pith/arXiv arXiv 2019

  25. [25]

    Harald Steck. 2018. Calibrated recommendations. InProceedings of the 12th ACM conference on recommender systems. 154–162

    2018

  26. [26]

    Fei Sun, Jun Liu, Jian Wu, Changhua Pei, Xiao Lin, Wenwu Ou, and Peng Jiang

  27. [27]

    Yi Tay, Mostafa Dehghani, Dara Bahri, and Donald Metzler. 2020. Efficient Transformers: A Survey.arXiv preprint arXiv:2009.06732(2020)

    Pith/arXiv arXiv 2020

  28. [28]

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need.Advances in neural information processing systems30 (2017)

    2017

  29. [29]

    Hao Wang, Wei Guo, Luankang Zhang, Jin Yao Chin, Yufei Ye, Huifeng Guo, Yong Liu, Defu Lian, Ruiming Tang, and Enhong Chen. 2025. Generative large recom- mendation models: Emerging trends in llms for recommendation. InCompanion Proceedings of the ACM on Web Conference 2025. 49–52

    2025

  30. [30]

    Ruoxi Wang, Bin Fu, Gang Fu, and Mingliang Wang. 2017. Deep & cross network for ad click predictions. InProceedings of the ADKDD’17. 1–7

    2017

  31. [31]

    Ruoxi Wang, Rakesh Shivanna, Derek Cheng, Sagar Jain, Dong Lin, Lichan Hong, and Ed Chi. 2021. Dcn v2: Improved deep & cross network and practical lessons for web-scale learning to rank systems. InProceedings of the web conference 2021. 1785–1797

    2021

  32. [32]

    Xue Xia, Pong Eksombatchai, Nikil Pancha, Dhruvil Deven Badani, Po-Wei Wang, Neng Gu, Saurabh Vishwas Joshi, Nazanin Farahpour, Zhiyuan Zhang, and An- drew Zhai. 2023. Transact: Transformer-based realtime user action model for recommendation at pinterest. InProceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 5249–5259

    2023

  33. [33]

    Xue Xia, Saurabh Joshi, Kousik Rajesh, Kangnan Li, Yangyi Lu, Nikil Pancha, Dhruvil Badani, Jiajing Xu, and Pong Eksombatchai. 2025. TransAct V2: Lifelong User Action Sequence Modeling on Pinterest Recommendation. InProceedings of the 34th ACM International Conference on Information and Knowledge Management. 6881–6882

    2025

  34. [34]

    Zhichen Zeng, Xiaolong Liu, Mengyue Hang, Xiaoyi Liu, Qinghai Zhou, Chaofei Yang, Yiqun Liu, Yichen Ruan, Laming Chen, Yuxin Chen, et al. 2025. InterFormer: Effective Heterogeneous Interaction Learning for Click-Through Rate Prediction. InProceedings of the 34th ACM International Conference on Information and Knowledge Management. 6225–6233

    2025

  35. [35]

    Jiaqi Zhai, Lucy Liao, Xing Liu, Yueming Wang, Rui Li, Xuan Cao, Leon Gao, Zhao- jie Gong, Fangda Gu, Michael He, et al. 2024. Actions speak louder than words: Trillion-parameter sequential transducers for generative recommendations.arXiv preprint arXiv:2402.17152(2024)

    Pith/arXiv arXiv 2024

  36. [36]

    Shuai Zhang, Lina Yao, Aixin Sun, and Yi Tay. 2019. Deep learning based recom- mender system: A survey and new perspectives.ACM computing surveys (CSUR) 52, 1 (2019), 1–38

    2019

  37. [37]

    Zhaoqi Zhang, Haolei Pei, Jun Guo, Tianyu Wang, Yufei Feng, Hui Sun, Shaowei Liu, and Aixin Sun. 2025. OneTrans: Unified Feature Interaction and Sequence Modeling with One Transformer in Industrial Recommender.arXiv preprint arXiv:2510.26104(2025)

    arXiv 2025

  38. [38]

    Guorui Zhou, Xiaoqiang Zhu, Chenru Song, Ying Fan, Han Zhu, Xiao Ma, Yanghui Yan, Junqi Jin, Han Li, and Kun Gai. 2018. Deep interest network for click-through rate prediction. InProceedings of the 24th ACM SIGKDD international conference on knowledge discovery & data mining. 1059–1068

    2018

  39. [39]

    Jie Zhu, Zhifang Fan, Xiaoxie Zhu, Yuchen Jiang, Hangyu Wang, Xintian Han, Haoran Ding, Xinmin Wang, Wenlin Zhao, Zhen Gong, et al. 2025. Rankmixer: Scaling up ranking models in industrial recommenders. InProceedings of the 34th ACM International Conference on Information and Knowledge Management. 6309–6316. KDD ’26, August 09–13, 2026, Jeju Island, Repub...

    2025

  40. [2016]

    InProceedings of the 1st workshop on deep learning for recommender systems

    Wide & deep learning for recommender systems. InProceedings of the 1st workshop on deep learning for recommender systems. 7–10

  41. [2019]

    InProceedings of the 28th ACM international conference on information and knowledge management

    BERT4Rec: Sequential recommendation with bidirectional encoder rep- resentations from transformer. InProceedings of the 28th ACM international conference on information and knowledge management. 1441–1450

  42. [2023]

    Recommender systems with generative retrieval.Advances in Neural Information Processing Systems36 (2023), 10299–10315

    2023