arXiv subjects

Ruoyan Wang

Publications and source records attributed to Ruoyan Wang.

5 recordsLinked to original sources

arXiv2026

CADET: Context-Conditioned Ads CTR Prediction With a Decoder-Only Transformer

Click-through rate (CTR) prediction is fundamental to online advertising systems. While Deep Learning Recommendation Models (DLRMs) with explicit feature interactions have long dominated this domain, recent advances in generative recommenders have shown promising results in content recommendation. However, adapting these transformer-based architectures to ads CTR prediction still presents unique challenges, including handling post-scoring contextual signals, maintaining offline-online consistency, and scaling to industrial workloads. We present CADET (Context-Conditioned Ads Decoder-Only Transformer), an end-to-end decoder-only transformer for ads CTR prediction deployed at LinkedIn. Our approach introduces several key innovations: (1) a context-conditioned decoding architecture with multi-tower prediction heads that explicitly model post-scoring signals such as ad position, resolving the chicken-and-egg problem between predicted CTR and ranking; (2) a self-gated attention mechanism that stabilizes training by adaptively regulating information flow at both representation and interaction levels; (3) a timestamp-based variant of Rotary Position Embedding (RoPE) that captures temporal relationships across timescales from seconds to months; (4) session masking strategies that prevent the model from learning dependencies on unavailable in-session events, addressing train-serve skew; and (5) production engineering techniques including tensor packing, sequence chunking, and custom Flash Attention kernels that enable efficient training and serving at scale. In online A/B testing, CADET achieves a 11.04\% CTR lift compared to the production LiRank baseline model, a hybrid ensemble of DCNv2 and sequential encoders. The system has been successfully deployed on LinkedIn's advertising platform, serving the main traffic for homefeed sponsored updates.

cs.LG↗

arXiv2024

LiRank: Industrial Large Scale Ranking Models at LinkedIn

We present LiRank, a large-scale ranking framework at LinkedIn that brings to production state-of-the-art modeling architectures and optimization methods. We unveil several modeling improvements, including Residual DCN, which adds attention and residual connections to the famous DCNv2 architecture. We share insights into combining and tuning SOTA architectures to create a unified model, including Dense Gating, Transformers and Residual DCN. We also propose novel techniques for calibration and describe how we productionalized deep learning based explore/exploit methods. To enable effective, production-grade serving of large ranking models, we detail how to train and compress models using quantization and vocabulary compression. We provide details about the deployment setup for large-scale use cases of Feed ranking, Jobs Recommendations, and Ads click-through rate (CTR) prediction. We summarize our learnings from various A/B tests by elucidating the most effective technical approaches. These ideas have contributed to relative metrics improvements across the board at LinkedIn: +0.5% member sessions in the Feed, +1.76% qualified job applications for Jobs search and recommendations, and +4.3% for Ads CTR. We hope this work can provide practical insights and solutions for practitioners interested in leveraging large-scale deep ranking systems.

cs.LG↗

arXiv2023

Detection of the infrared aurora at Uranus with Keck-NIRSPEC

Near infrared (NIR) wavelength observations of Uranus have been unable to locate any infrared aurorae, despite many attempts to do so since the 1990s. While at Jupiter and Saturn, NIR investigations have redefined our understanding of magnetosphere ionosphere thermosphere coupling, the lack of NIR auroral detection at Uranus means that we have lacked a window through which to study these processes at Uranus. Here we present NIR Uranian observations with the Keck II telescope taken on the 5 September 2006 and detect enhanced $\text{H}_{\text{3}}^{\text{+}}$ emissions. Analysing temperatures and column densities, we identify an 88\% increase in localized $\text{H}_{\text{3}}^{\text{+}}$ column density, with no significant temperature increases, consistent with auroral activity generating increased ionization. By comparing these structures against the $\text{Q}_{\text{3}}^{\text{mp}}$ magnetic field model and the Voyager 2 ultraviolet observations, we suggest that these regions make up sections of the northern aurora.

astro-ph.EP↗

arXiv2021

The Sonora Brown Dwarf Atmosphere and Evolution Models I. Model Description and Application to Cloudless Atmospheres in Rainout Chemical Equilibrium

We present a new generation of substellar atmosphere and evolution models, appropriate for application to studies of L, T, and Y-type brown dwarfs and self-luminous extrasolar planets. The atmosphere models describe the expected temperature-pressure profiles and emergent spectra of atmospheres in radiative-convective equilibrium with effective temperatures and gravities within the ranges $200\le T_{\rm eff}\le2400\,\rm K$ and $2.5\le \log g \le 5.5$. These ranges encompass masses from about 0.5 to 85 Jupiter masses for a set of metallicities ($[{\rm M/H}] = -0.5$ to $+0.5$), C/O ratios (from 0.5 to 1.5 times that of solar), and ages. The evolution tables describe the cooling of these substellar objects through time. These models expand the diversity of model atmospheres currently available, notably to cooler effective temperatures and greater ranges in C/O. Notable improvements from past such models include updated opacities and atmospheric chemistry. Here we describe our modeling approach and present our initial tranche of models for cloudless, chemical equilibrium atmospheres. We compare the modeled spectra, photometry, and evolution to various datasets.

astro-ph.SR↗

arXiv2015

Theano-based Large-Scale Visual Recognition with Multiple GPUs

In this report, we describe a Theano-based AlexNet (Krizhevsky et al., 2012) implementation and its naive data parallelism on multiple GPUs. Our performance on 2 GPUs is comparable with the state-of-art Caffe library (Jia et al., 2014) run on 1 GPU. To the best of our knowledge, this is the first open-source Python-based AlexNet implementation to-date.

cs.LG↗