Pith. sign in

REVIEW 9 cited by

RankMixer: Scaling Up Ranking Models in Industrial Recommenders

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2507.15551 v3 pith:ZDSE6WYL submitted 2025-07-21 cs.IR

RankMixer: Scaling Up Ranking Models in Industrial Recommenders

classification cs.IR
keywords rankmixermodelmodelsrankingscalingservingcosthigh
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Recent progress on large language models (LLMs) has spurred interest in scaling up recommendation systems, yet two practical obstacles remain. First, training and serving cost on industrial Recommenders must respect strict latency bounds and high QPS demands. Second, most human-designed feature-crossing modules in ranking models were inherited from the CPU era and fail to exploit modern GPUs, resulting in low Model Flops Utilization (MFU) and poor scalability. We introduce RankMixer, a hardware-aware model design tailored towards a unified and scalable feature-interaction architecture. RankMixer retains the transformer's high parallelism while replacing quadratic self-attention with multi-head token mixing module for higher efficiency. Besides, RankMixer maintains both the modeling for distinct feature subspaces and cross-feature-space interactions with Per-token FFNs. We further extend it to one billion parameters with a Sparse-MoE variant for higher ROI. A dynamic routing strategy is adapted to address the inadequacy and imbalance of experts training. Experiments show RankMixer's superior scaling abilities on a trillion-scale production dataset. By replacing previously diverse handcrafted low-MFU modules with RankMixer, we boost the model MFU from 4.5\% to 45\%, and scale our ranking model parameters by 100x while maintaining roughly the same inference latency. We verify RankMixer's universality with online A/B tests across two core application scenarios (Recommendation and Advertisement). Finally, we launch 1B Dense-Parameters RankMixer for full traffic serving without increasing the serving cost, which improves user active days by 0.3\% and total in-app usage duration by 1.08\%.

discussion (0)

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

Forward citations

Cited by 9 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Selective Test-Time Compute Scaling for Click-Through Rate Prediction via Uncertainty-Triggered Feature Path Exploration

    cs.LG 2026-05 unverdicted novelty 7.0

    UTTSI selectively scales test-time compute for CTR prediction by triggering stochastic feature-path exploration only on high-uncertainty instances, yielding gains on four datasets and a 5.3% online CTR lift.

  2. TokenFormer: Unify the Multi-Field and Sequential Recommendation Worlds

    cs.IR 2026-04 unverdicted novelty 7.0

    TokenFormer unifies multi-field and sequential recommendation modeling via bottom-full-top-sliding attention and non-linear interaction representations to avoid sequential collapse and deliver state-of-the-art performance.

  3. Tencent Advertising Algorithm Challenge 2025: All-Modality Generative Recommendation

    cs.IR 2026-04 accept novelty 7.0

    Releases TencentGR-1M and TencentGR-10M datasets with baselines for all-modality generative recommendation in advertising, including weighted evaluation for conversions.

  4. Make It Long, Keep It Fast: End-to-End 10K Long User Behavior Sequence Modeling for Billion-Scale Douyin Recommendation

    cs.LG 2025-11 conditional novelty 6.0

    Douyin deploys stacked target-to-history cross attention and request-level batching to scale end-to-end recommendation modeling to 10k-length histories, observing scaling-law gains and live engagement improvements.

  5. A Survey on Generative Recommendation: Data, Model, and Tasks

    cs.IR 2025-10 accept novelty 6.0

    This survey organizes generative recommendation into data, model, and task dimensions, identifying five advantages including world knowledge integration and creative generation while noting challenges in benchmarks an...

  6. Make It Long, Keep It Fast: End-to-End 10K Long User Behavior Sequence Modeling for Billion-Scale Douyin Recommendation

    cs.LG 2025-11 unverdicted novelty 5.0

    Introduces STCA for linear-complexity target-to-history attention, RLB for shared user encoding across targets, and length-extrapolative training to enable end-to-end 10K sequence modeling with observed scaling-law ga...

  7. OneSearch-V2: The Latent Reasoning Enhanced Self-distillation Generative Search Framework

    cs.IR 2026-03 unverdicted novelty 4.0

    OneSearch-V2 improves generative retrieval via latent reasoning and self-distillation, achieving +3.98% item CTR, +2.07% buyer volume, and +2.11% order volume in online A/B tests.

  8. Joint Model Parameter Scaling and Universal-Domain Data Integration for E-commerce Search Ranking

    cs.IR 2026-03 unverdicted novelty 4.0

    UniScale couples entire-space data construction with a hierarchical fusion transformer to improve scaling behavior and deliver 1.70% purchase and 2.04% GMV lifts in large-scale e-commerce search A/B tests.

  9. On the Practice of Scaling Search Conversion Rate Prediction

    cs.IR 2026-05 unverdicted novelty 2.0

    Empirical scaling of backbone, embeddings, and data shows largely independent additive gains, enabling a deployed model with 2.5x data and 8x compute that delivers +2.6% CVR improvement with minimal latency change.