Pith. sign in

REVIEW 4 major objections 6 minor 23 references

Progressive Alignment of Recommender Foundation Model through Multi-Phase Post-Training

T0 review · 4 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read The paper claims that a three-phase post-training sequence—linear probing, full fine-tuning, then GRPO-based reinforcement fine-tuning against a learned reward model—beats single-phase SFT on click and funnel-depth ranking, and that the…

desk verdict A useful industrial recipe with real online A/B evidence; the RFT stage is formally underspecified and offline stats are thin. read the letter →

arxiv 2608.06792 v1 pith:4QVE5DUN submitted 2026-08-07 cs.IR cs.AI

Oseong Choi , Hoeinn Kim , Jihoon Lee , Byungsoo Kang , Taeyeong Jang This is my paper · ORCID
classification cs.IRcs.AI
keywords recommendersystemsfoundationmodelsprogressivepost-trainingreinforcementfine-tuningrewardmodelingGRPODPOordinalregression
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper proposes a three-phase post-training recipe for adapting a pretrained recommender foundation model to a concrete serving surface. Phase one freezes the backbone and trains only the new downstream heads (linear probing, LP); phase two unfreezes everything and fine-tunes jointly, with a smaller learning rate on the backbone (full fine-tuning, FFT); phase three runs GRPO-style reinforcement fine-tuning against a reward model learned from a six-level engagement funnel, with DPO as an alternative. The paper tries to establish that this progression is more stable and more accurate than single-phase SFT, and that using the reward model as a training signal beats using it directly as the serving ranker. If true, the payoff is a practical path from a shared sequential user-behavior model to a deployed policy that ranks for immediate clicks and for deeper business outcomes at the same time.

What carries the argument

The engine is the three-phase pipeline itself plus the reward model. The reward model maps sparse, delayed business outcomes into dense scalar utility by ordinal regression over six funnel stages, with a clipped inverse-propensity weight and self-normalization to correct exposure bias. Phase 3 wraps that reward in GRPO: for each user state the current policy greedily selects top-32 titles, the reward model scores them, advantages are normalized within the group, and the policy is updated with a clipped surrogate objective plus a KL penalty against the frozen post-FFT reference policy. The same reward scores can also be converted into preference pairs for a DPO variant. The load-bearing separation is that dense implicit feedback trains the policy, sparse business labels train the reward, and the reward steers the policy.

What would settle it

Compute $\sum_{a\in\mathcal{V}} \pi_\theta(a|u)$ for a trained policy and compare greedy top-32 selection with sampling from the stated distribution. If the policy scores are not a normalized distribution over the full catalog, or if replacing the learned reward model with a deterministic oracle or with observed funnel labels does not change the offline gains, then the reported improvements cannot be attributed to GRPO-based alignment.

Watch

Extended reading notes

Core claim

The central discovery, on the paper's own terms, is that the business-metric reward and the serving policy should play different roles. A policy head trained on dense implicit feedback (clicks) provides full-catalog discrimination: in the offline ablation it reaches Rank NDCG 0.437 after LP to FFT. The reward model, trained with inverse-propensity-weighted ordinal regression on a six-level funnel, captures engagement depth: used directly as a ranker it reaches Funnel NDCG 0.637 but Rank NDCG only 0.394. GRPO fine-tuning of the click-trained policy against this reward model takes the policy to Rank NDCG 0.463 and Funnel NDCG 0.637, matching the reward model's funnel quality while retaining and exceeding the click-ranking discrimination. The paper reads this as evidence that sparse business objectives are best used as alignment signals, not as supervised labels or direct serving scores.

Load-bearing premise

The GRPO stage assumes that $\pi_\theta(a|u)$ is a genuine probability distribution over candidate items, so that policy ratios, advantage normalization, and the KL penalty in Eq. (1) describe an actual stochastic policy; the paper never defines that distribution and Algorithm 1 selects the top-32 items greedily, so the reinforcement objective may not be the GRPO procedure it is claimed to be.

Editorial extensions

If this is right

  • A pretrained recommender foundation model can be introduced into an existing production ranking stack through LP to FFT and yield consistent online gains over a conventional CTR/CVR product-rank baseline, even before any reinforcement stage.
  • Reward-model alignment produces a serving policy whose funnel-depth ranking matches the reward model itself while retaining much stronger full-catalog click ranking, so business objectives can be injected without sacrificing ranking discrimination.
  • GRPO with a learned reward model emphasizes mid-to-deep engagement (Engaged/Imp and LastFree/Imp), whereas DPO with observed funnel preferences emphasizes immediate clicks, so the choice of alignment method selects which part of the engagement funnel the policy optimizes.
  • The paper's stabilization choices—LP before FFT and a one-order-of-magnitude smaller learning rate on the backbone during FFT and RFT—are presented as necessary for the gains: single-phase LP or FFT alone underperforms the two-phase sequence on both reported metrics.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper does not test whether the gains persist when the policy is sampled stochastically rather than greedily top-32; a natural extension is to compare GRPO with full-catalog sampling against the greedy variant to see how much of the advantage comes from the objective versus the candidate-selection heuristic.
  • The six-level funnel is platform-specific, but the separation principle—train the serving policy on dense feedback, train the reward on sparse business outcomes, then align—should transfer to other marketplaces with a well-defined engagement ladder, such as video completion or purchase conversion.
  • The observation that the reward model underperforms on Rank NDCG when used directly suggests that business-utility scores are not calibrated for discrimination; an implicit extension is that calibrating or re-ranking reward scores might narrow the gap between serving the reward model and aligning with it.
  • The paper's online comparison is against a non-foundation production control, not against a foundation model aligned with other RLHF variants; an open question is how GRPO w/RM would fare against PPO or reward-model-free policy-gradient baselines on the same funnel metrics.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes a three-phase post-training pipeline for adapting a pretrained recommender foundation model to a specific serving surface. Phase 1 (LP) freezes the backbone and trains downstream heads; Phase 2 (FFT) unfreezes all parameters with a lower backbone learning rate; Phase 3 (RFT) applies GRPO or DPO using a reward model trained on a six-level engagement funnel with inverse propensity weighting. Offline experiments on a three-week production log report that the full pipeline (LP→FFT→GRPO) improves Rank NDCG and Funnel NDCG over one- and two-phase SFT baselines, and that aligning a click-trained policy with the reward model beats serving directly with the reward model. Online A/B tests report significant positive lifts over a non-foundation production control on in-session metrics.

Significance. If the method performs as reported, the paper's contribution is practically useful: it demonstrates a staged recipe for adapting a sequential recommender foundation model to a production surface, with an explicit separation between dense-signal adaptation and sparse business-metric alignment. The online A/B experiment against a conventional production model is a particularly valuable piece of evidence, and the reference implementation supports reproducibility. However, the central technical novelty—the RFT stage—is formally under-specified in the current manuscript, and the offline evaluation lacks the statistical rigor needed to support the claimed comparisons. The online results are encouraging, but the offline claims about the relative merits of GRPO/DPO and of reward-based alignment rest on undefined quantities and unquantified variance.

major comments (4)
  1. [Section 3.2.2, Eq. (1), Algorithm 1 lines 20–26] The policy ratio ρ_k(θ)=π_θ(a_k)/π_ref(a_k) and the KL term D_KL(π_θ(·|s)∥π_ref(·|s)) require π_θ(·|u) to be a normalized probability distribution over a fixed action set. The manuscript never defines this distribution. In Section 4.1 the serving policy head is described as predicting 'whether a click occurs after impression'—a binary event probability, not a distribution over candidate items. Algorithm 1 line 20 greedily selects top-K items 'using π_θ(·|u)', but if π_θ is a deterministic score, no stochastic policy exists; if π_θ is a softmax over the top-K set, the action set changes as the policy updates, so π_ref and π_θ are not defined on a common set; if π_θ is a softmax over the full catalog, the paper does not describe how the KL term is computed or how actions are sampled. Consequently, Eq. (1) is not a well-defined GRPO objective, and the gains attributed to 'GRPO w/ RM' in Table 1 cannot be unambiguously attributed to the stated algorithm. Please define the action distribution precisely (e.g., a softmax over a fixed candidate set with temperature) and adapt the algorithm and the KL term accordingly.
  2. [Section 4.1, Dataset] The test set consists of 10% of users sampled in the same three-week window used for training, and the text does not state whether these users' interactions were excluded from the training set. If they were not excluded, the offline metrics are potentially inflated by training/test overlap, and the differences between configurations in Table 1 are confounded. Please clarify the split (e.g., user-level holdout, session-level holdout) and explicitly state that no test user appears in training, or restructure the evaluation to avoid overlap.
  3. [Table 1] No variance, confidence interval, or significance test is reported for any offline comparison. For instance, the claimed advantage of 'GRPO w/ RM' (Rank NDCG 0.463) over the '2-Phase SFT' baseline (0.437) is presented as a conclusion without an error bar or p-value. Since the paper's central offline claims are pairwise comparisons among methods, please report standard errors or bootstrap confidence intervals (ideally across multiple training runs or user resamples) and, where possible, significance tests.
  4. [Section 4.1.1 'Funnel NDCG' and Section 3.1.2] The graded relevance labels used in Funnel NDCG are the same six-level funnel depths used to train the reward model via ordinal regression. The RFT methods in Table 1 are optimized against this reward model, so their Funnel NDCG improvements are partially self-referential: the metric evaluates ranking by the very labels that the alignment signal was trained to predict. The online A/B results provide external validation, but the offline claim that reward-based alignment is better than serving with the reward model itself should be supported by a metric whose labels are not used to train the reward model, or the circularity should be acknowledged and discussed.
minor comments (6)
  1. [Section 4.1] The HSTU backbone is not explicitly cited at the point of use; please add a citation to the appropriate reference (e.g., reference [19]) when introducing the HSTU encoder.
  2. [Section 3.2.2] The phrase 'the policy head and FM backbone is optimized' should read 'are optimized'.
  3. [Section 3.1.2] The equation for the self-normalized IPS weight is garbled in the manuscript text; please ensure it is typeset correctly (e.g., w̃_uv = s_uv / Σ_{(u',v')∈D_r} s_{u'v'}).
  4. [Section 4.2] The statistical test yielding 'p < 0.001' is not described; please state the test procedure and whether multiple-comparison corrections were applied across the four metrics.
  5. [Section 4.1 and Section 3.2.2] The notation π_θ is used both for the click-prediction head in Section 4.1 and for the policy in the GRPO objective in Section 3.2.2 without an explicit definitional link; please clarify the relationship (the same issue is raised in Major Comment 1, but a one-sentence clarification would help the reader).
  6. [Figure 3] The text refers to 'the daily trajectory of Click/Imp and LastPaid/Imp,' but the figure description does not mention error bars or confidence bands; please add them so that the claimed stability can be assessed visually.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity found: the paper's central comparisons are empirical and independently grounded, and no derivation reduces to its own inputs.

full rationale

The paper's main claims are supported by empirical offline and online comparisons rather than by a derivation that reduces to its inputs. The reward model in Section 3.1.2 is trained on logged funnel-depth labels, and Funnel NDCG in Section 4.1.1 assigns graded relevance from the same six-level engagement funnel; this overlap means the reward model and the metric share a target construct. However, the paper's headline claim that reward-based alignment yields a stronger serving policy than directly using the reward model for ranking is grounded in Rank NDCG, which uses click labels not contained in the reward model's training target. The RFT stage in Section 3.2.2 does not invoke any author-specific uniqueness theorem, and the LP-FFT motivation is supported by external citations ([7], [8]) rather than by a self-citation chain. The formal underspecification of the policy distribution π_θ(a|u) in Algorithm 1 lines 20–26, noted in the reader's take, is an internal-consistency or correctness concern, not circularity: it does not make any reported quantity equal to its own input by construction. No fitted parameter is renamed as a prediction, and no result is forced by definition or by the authors' prior work.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The framework is empirical, so the ledger is short. No new physical or conceptual entities are introduced; the reward model, propensity head, and policy modules are standard ML components. The uncharged load-bearing assumptions are the funnel utility model, the exposure-bias correction, the transferability of pretrained representations, and the implicit stochastic-policy interpretation of a deterministic ranker in RFT. The listed free parameters are hyperparameters of the schedule and RL objectives, none justified by sensitivity analysis.

free parameters (6)
  • tau_IPS inverse propensity clipping threshold = not reported
    Used in Section 3.1.2 to cap self-normalized inverse propensity weights; the value controls the bias-variance tradeoff of the reward model and is not stated.
  • GRPO KL coefficient beta = 0.001
    Set in Section 4.1 to constrain policy drift from the reference policy; no sensitivity analysis is provided.
  • GRPO surrogate clip epsilon = 0.1
    PPO-style clipping bound used in the RFT objective; chosen without sensitivity analysis.
  • Candidate group size K = 32
    Top-32 greedy candidates per user state, chosen to match the serving setting; the impact of K is not studied.
  • Learning rates = LP 1e-3, FFT 1e-4, FM 1e-5, RFT policy 1e-5, RFT FM 1e-6
    Discriminative learning-rate control is a load-bearing stability mechanism in Phases 2 and 3; the values are hand-set.
  • Epoch allocation = 50 LP + 50 FFT + 50 RFT
    Training budget is split across phases; no ablation of the split is reported.
assumptions (4)
  • domain assumption The six-level engagement funnel (impression, click, initial episode, free completion, paid entry, paid completion) is a valid ordinal representation of business utility.
    Section 3.1.2 defines the reward target as the deepest funnel stage; if the stages are not monotone in utility or are misordered, the reward model is misspecified.
  • domain assumption Logged feedback is subject to exposure bias that can be corrected by a learned propensity head with clipped self-normalized inverse propensity scores.
    Section 3.1.2 introduces the propensity head and SNIPS weights; no propensity model architecture, validation, or clipping value is given.
  • ad hoc to paper The GRPO and DPO objectives are well-defined for the serving policy, which requires a stochastic policy distribution over items.
    Algorithm 1 selects top-K items greedily, and the paper never defines pi_theta(a|u) as a probability distribution or a sampling procedure; without this assumption the policy ratio and KL term have no clear meaning.
  • domain assumption Pretrained HSTU representations transfer to the downstream discovery task and are worth preserving through discriminative learning rates.
    Section 3.2.1 and the offline experiments assume the FM backbone provides general-purpose behavioral representations that should not be destabilized.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Progressive Alignment of Recommender Foundation Model through Multi-Phase Post-Training." pith.science (2026). https://pith.science/paper/4QVE5DUN

@misc{pith2026260806792,
  author       = {Pith},
  title        = {Pith review of: Progressive Alignment of Recommender Foundation Model through Multi-Phase Post-Training},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4QVE5DUN}},
  note         = {Machine review of arXiv:2608.06792}
}
read the original abstract

Foundation model(FM) for recommendation has shown strong ability to model long-horizon sequential user behavior. In practice, a single pretrained foundation model is often adapted to diverse downstream serving surfaces through Supervised Fine-Tuning(SFT). However, optimizing task-specific objectives such as clicks or likes does not necessarily align the serving policy with the business metrics that determine recommendation quality. We propose a three-phase progressive post-training framework that explicitly separates downstream adaptation from business-metric alignment. The adaptation stage is decomposed into Linear Probing(LP) and Full Fine-Tuning(FFT): LP first stabilizes randomly initialized downstream heads within a frozen pretrained representation space, and FFT then jointly specializes the full model for the target task. On top of this stabilized policy, Reinforcement Fine-Tuning(RFT) aligns the model with practical business objectives using a learned reward model. Rather than directly optimizing the serving policy on sparse business targets, we train the policy on dense implicit feedback and use business-metric supervision only for reward modeling. Offline experiments show that the progressive LP-FFT-RFT framework outperforms single-phase alternatives, and that reward-based alignment yields a stronger serving policy than directly using the reward model itself for ranking. Large-scale online A/B tests further show that the proposed framework improves production recommendation quality over a conventional non-foundation baseline. A reference implementation is available at https://github.com/webtoon/rec-fm-progressive-alignment

Figures

Figures reproduced from arXiv: 2608.06792 by the authors.

Figure 1
Figure 1. Overall architecture of the foundation–expert [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Three-phase progressive post-training pipeline for foundation recommenders. Solid lines indicate gradient flow and [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Relative online lift over the production control model across in-session metrics. The top plot shows difference-in [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

23 extracted references · 5 canonical work pages

  1. [1]

    Maarten Buyl, Paul Missault, and Pierre-Antoine Sondag. 2023. RankFormer: Listwise Learning-to-Rank Using Listwide Labels. InProceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD). ACM, 3762–3773. doi:10.1145/3580305.3599892

    arXiv 2023

  2. [2]

    Xiangyi Chen, Kousik Rajesh, Matthew Lawhon, Zelun Wang, Hanyu Li, Haomiao Li, Saurabh Vishwas Joshi, Pong Eksombatchai, Jaewon Yang, Yi-Ping Hsu, Jiajing Xu, and Charles Rosenberg. 2025. PinFM: Foundation Model for User Activity Sequences at a Billion-scale Visual Discovery Platform. InProceedings of the Nineteenth ACM Conference on Recommender Systems (...

    arXiv 2025

  3. [3]

    Yuqi Gong, Xichen Ding, Yehui Su, Kaiming Shen, Zhongyi Liu, and Guannan Zhang. 2023. An Unified Search and Recommendation Foundation Model for Cold- Start Scenario. InProceedings of the ACM International Conference on Information and Knowledge Management (CIKM). ACM, 4595–4601. doi:10.1145/3583780. 3614657

    doi:10.1145/3583780 2023

  4. [4]

    Balázs Hidasi, Alexandros Karatzoglou, Linas Baltrunas, and Domonkos Tikk

  5. [5]

    Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen

    Edward J. Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2022. LoRA: Low-Rank Adaptation of Large Language Models. InInternational Conference on Learning Representations (ICLR)

    2022

  6. [6]

    Wang-Cheng Kang and Julian McAuley. 2018. Self-Attentive Sequential Rec- ommendation. InIEEE International Conference on Data Mining (ICDM). IEEE Computer Society, 197–206. doi:10.1109/ICDM.2018.00035

    arXiv 2018

  7. [7]

    Ananya Kumar, Aditi Raghunathan, Robbie Jones, Tengyu Ma, and Percy Liang

    work page

  8. [8]

    2025.Re- inforcement Fine-Tuning Naturally Mitigates Forgetting in Continual Post-Training

    Song Lai, Haohan Zhao, Rong Feng, Changyi Ma, Wenzhuo Liu, Hongbo Zhao, Xi Lin, Dong Yi, Qingfu Zhang, Hongbin Liu, Gaofeng Meng, and Fei Zhu. 2025.Re- inforcement Fine-Tuning Naturally Mitigates Forgetting in Continual Post-Training. arXiv:2507.05386

    arXiv 2025

Show all 23 references
  1. [9]

    2025.Realizing Scaling Laws in Recommender Systems: A Foundation-Expert Paradigm for Hyperscale Model Deployment

    Dai Li, Kevin Course, Wei Li, Hongwei Li, Jie Hua, Yiqi Chen, Zhao Zhu, Rui Jian, Xuan Cao, Bi Xue, Yu Shi, Jing Qian, Kai Ren, Matt Ma, Qunshu Zhang, and Rui Li. 2025.Realizing Scaling Laws in Recommender Systems: A Foundation-Expert Paradigm for Hyperscale Model Deployment. ...

    2025 arXiv

  2. [10]

    Mingfu Liang, Xi Liu, Rong Jin, Boyang Liu, Qiuling Suo, Qinghai Zhou, Song Zhou, Laming Chen, Hua Zheng, Zhiyuan Li, Shali Jiang, Jiyan Yang, Xiaozhen Xia, Fan Yang, Yasmine Badr, Ellie Wen, Shuyu Xu, Hansey Chen, Zhengyu Zhang, Jade Nie, Chunzhi Yang, Zhichen Zeng, Weilin Zh...

    2025

  3. [11]

    Yun Luo, Zhen Yang, Fandong Meng, Yafu Li, Jie Zhou, and Yue Zhang. 2025. An Empirical Study of Catastrophic Forgetting in Large Language Models Dur- ing Continual Fine-tuning.IEEE Transactions on Audio, Speech and Language Processing33 (2025), 3776–3786. doi:10.1109/TASLPRO.2...

    2025

  4. [12]

    Long Ouyang, Jeff Wu, Xu Jiang, Diogo Almeida, Carroll L. Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, John Schul- man, Jacob Hilton, Fraser Kelton, Luke Miller, Maddie Simens, Amanda Askell, Peter Welinder, Paul F. Christiano, Jan Leike...

    2022

  5. [13]

    Manning, Stefano Ermon, and Chelsea Finn

    Rafael Rafailov, Archit Sharma, Eric Mitchell, Christopher D. Manning, Stefano Ermon, and Chelsea Finn. 2023. Direct Preference Optimization: Your Language Model is Secretly a Reward Model. InAdvances in Neural Information Processing Systems (NeurIPS), Vol. 36. 53728–53741

    2023

  6. [14]

    Keshavan, Trung Vu, Lukasz Heldt, Lichan Hong, Yi Tay, Vinh Q

    Shashank Rajput, Nikhil Mehta, Anima Singh, Raghunandan H. Keshavan, Trung Vu, Lukasz Heldt, Lichan Hong, Yi Tay, Vinh Q. Tran, Jonah Samost, Maciej Kula, Ed H. Chi, and Maheswaran Sathiamoorthy. 2023. Recommender Systems with Generative Retrieval. InAdvances in Neural Informa...

    2023

  7. [15]

    Ramasesh, Ethan Dyer, and Maithra Raghu

    Vinay V. Ramasesh, Ethan Dyer, and Maithra Raghu. 2021. Anatomy of Cata- strophic Forgetting: Hidden Representations and Task Semantics. InInternational Conference on Learning Representations (ICLR)

    2021

  8. [16]

    2017.Proximal Policy Optimization Algorithms

    John Schulman, Filip Wolski, Prafulla Dhariwal, Alec Radford, and Oleg Klimov. 2017.Proximal Policy Optimization Algorithms. arXiv:1707.06347

    2017 arXiv

  9. [17]

    Zhihong Shao, Peiyi Wang, Qihao Zhu, Runxin Xu, Junxiao Song, Xiao Bi, Haowei Zhang, Mingchuan Zhang, Y. K. Li, Y. Wu, and Daya Guo. 2024.DeepSeek- Math: Pushing the Limits of Mathematical Reasoning in Open Language Models. arXiv:2402.03300

    2024 arXiv

  10. [18]

    Hao Wang, Tai-Wei Chang, Tianqiao Liu, Jianmin Huang, Zhichao Chen, Chao Yu, Ruopeng Li, and Wei Chu. 2022. ESCM2: Entire Space Counterfactual Multi- Task Model for Post-Click Conversion Rate Estimation. InProceedings of the 45th International ACM SIGIR Conference on Research ...

    2022

  11. [19]

    Jiaqi Zhai, Lucy Liao, Xing Liu, Yueming Wang, Rui Li, Xuan Cao, Leon Gao, Zhaojie Gong, Fangda Gu, Michael He, Yinghai Lu, and Yu Shi. 2024. Actions Speak Louder than Words: Trillion-Parameter Sequential Transducers for Gen- erative Recommendations. InProceedings of the 41st ...

    2024

  12. [20]

    Gaowei Zhang, Yupeng Hou, Hongyu Lu, Yu Chen, Wayne Xin Zhao, and Ji- Rong Wen. 2024. Scaling Law of Large Sequential Recommendation Models. In Proceedings of the 18th ACM Conference on Recommender Systems (RecSys). ACM, 444–453. doi:10.1145/3640457.3688129

    2024

  13. [21]

    2025.Large Foundation Model for Ads Recommendation

    Shangyu Zhang, Shijie Quan, Zhongren Wang, Junwei Pan, Tianqu Zhuang, Bo Fu, Yilong Sun, Jieying Lin, Jushuo Chen, Xiaotian Li, Zhixiang Feng, Xian Hu, Huiting Deng, Hua Lu, Jinpeng Wang, Boqi Dai, Xiaoyu Chen, Bin Hu, Lili Huang, Yanwen Wu, Yeshou Cai, Qi Zhou, Huang Tang, Ch...

    2025 arXiv

  14. [2016]

    In International Conference on Learning Representations (ICLR)

    Session-based Recommendations with Recurrent Neural Networks. In International Conference on Learning Representations (ICLR)

  15. [2022]

    InInternational Conference on Learning Representations (ICLR)

    Fine-Tuning Can Distort Pretrained Features and Underperform Out-of- Distribution. InInternational Conference on Learning Representations (ICLR)

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.