arXiv subjects

Yujing Zhang

Publications and source records attributed to Yujing Zhang.

33 records · Page 2Linked to original sources

arXiv2025

Gemini: A Family of Highly Capable Multimodal Models

This report introduces a new family of multimodal models, Gemini, that exhibit remarkable capabilities across image, audio, video, and text understanding. The Gemini family consists of Ultra, Pro, and Nano sizes, suitable for applications ranging from complex reasoning tasks to on-device memory-constrained use-cases. Evaluation on a broad range of benchmarks shows that our most-capable Gemini Ultra model advances the state of the art in 30 of 32 of these benchmarks - notably being the first model to achieve human-expert performance on the well-studied exam benchmark MMLU, and improving the state of the art in every one of the 20 multimodal benchmarks we examined. We believe that the new capabilities of the Gemini family in cross-modal reasoning and language understanding will enable a wide variety of use cases. We discuss our approach toward post-training and deploying Gemini models responsibly to users through services including Gemini, Gemini Advanced, Google AI Studio, and Cloud Vertex AI.

cs.CL↗

arXiv2025

Expectations, Explanations, and Embodiment: Attempts at Robot Failure Recovery

Expectations critically shape how people form judgments about robots, influencing whether they view failures as minor technical glitches or deal-breaking flaws. This work explores how high and low expectations, induced through brief video priming, affect user perceptions of robot failures and the utility of explanations in HRI. We conducted two online studies ($N=600$ total participants); each replicated two robots with different embodiments, Furhat and Pepper. In our first study, grounded in expectation theory, participants were divided into two groups, one primed with positive and the other with negative expectations regarding the robot's performance, establishing distinct expectation frameworks. This validation study aimed to verify whether the videos could reliably establish low and high-expectation profiles. In the second study, participants were primed using the validated videos and then viewed a new scenario in which the robot failed at a task. Half viewed a version where the robot explained its failure, while the other half received no explanation. We found that explanations significantly improved user perceptions of Furhat, especially when participants were primed to have lower expectations. Explanations boosted satisfaction and enhanced the robot's perceived expressiveness, indicating that effectively communicating the cause of errors can help repair user trust. By contrast, Pepper's explanations produced minimal impact on user attitudes, suggesting that a robot's embodiment and style of interaction could determine whether explanations can successfully offset negative impressions. Together, these findings underscore the need to consider users' expectations when tailoring explanation strategies in HRI. When expectations are initially low, a cogent explanation can make the difference between dismissing a failure and appreciating the robot's transparency and effort to communicate.

cs.RO↗

arXiv2024

Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context

In this report, we introduce the Gemini 1.5 family of models, representing the next generation of highly compute-efficient multimodal models capable of recalling and reasoning over fine-grained information from millions of tokens of context, including multiple long documents and hours of video and audio. The family includes two new models: (1) an updated Gemini 1.5 Pro, which exceeds the February version on the great majority of capabilities and benchmarks; (2) Gemini 1.5 Flash, a more lightweight variant designed for efficiency with minimal regression in quality. Gemini 1.5 models achieve near-perfect recall on long-context retrieval tasks across modalities, improve the state-of-the-art in long-document QA, long-video QA and long-context ASR, and match or surpass Gemini 1.0 Ultra's state-of-the-art performance across a broad set of benchmarks. Studying the limits of Gemini 1.5's long-context ability, we find continued improvement in next-token prediction and near-perfect retrieval (>99%) up to at least 10M tokens, a generational leap over existing models such as Claude 3.0 (200k) and GPT-4 Turbo (128k). Finally, we highlight real-world use cases, such as Gemini 1.5 collaborating with professionals on completing their tasks achieving 26 to 75% time savings across 10 different job categories, as well as surprising new capabilities of large language models at the frontier; when given a grammar manual for Kalamang, a language with fewer than 200 speakers worldwide, the model learns to translate English to Kalamang at a similar level to a person who learned from the same content.

cs.CL↗

arXiv2024

Enhancing Taobao Display Advertising with Multimodal Representations: Challenges, Approaches and Insights

Despite the recognized potential of multimodal data to improve model accuracy, many large-scale industrial recommendation systems, including Taobao display advertising system, predominantly depend on sparse ID features in their models. In this work, we explore approaches to leverage multimodal data to enhance the recommendation accuracy. We start from identifying the key challenges in adopting multimodal data in a manner that is both effective and cost-efficient for industrial systems. To address these challenges, we introduce a two-phase framework, including: 1) the pre-training of multimodal representations to capture semantic similarity, and 2) the integration of these representations with existing ID-based models. Furthermore, we detail the architecture of our production system, which is designed to facilitate the deployment of multimodal representations. Since the integration of multimodal representations in mid-2023, we have observed significant performance improvements in Taobao display advertising system. We believe that the insights we have gathered will serve as a valuable resource for practitioners seeking to leverage multimodal data in their systems.

cs.IR↗

arXiv2024

Decoupling Many-Body Interactions in CeO2 (111) Oxygen Vacancy Structure: Insights from Machine-Learning and Cluster Expansion

Oxygen vacancies (VO's) are of paramount importance in influencing the properties and applications of ceria (CeO2). Yet, comprehending the distribution and nature of the VO's poses a significant challenge due to the vast number of electronic configurations and intricate many-body interactions among VO's and polarons (Ce3+'s). In this study, we employed a combination of LASSO regression in machine learning, in conjunction with a cluster expansion model and first-principles calculations to decouple the interactions among the Ce3+'s and VO's, thereby circumventing the limitations associated with sampling electronic configurations. By separating these interactions, we identified specific electronic configurations characterized by the most favorable VO-Ce3+ attractions and the least Ce3+-Ce3+/VO-VO repulsions, which are crucial in determining the stability of vacancy structures. Through more than 10^8 Metropolis Monte Carlo samplings of Vo's and Ce3+ in the near-surface of CeO2(111), we explored potential configurations within an 8x8 supercell. Our findings revealed that oxygen vacancies tend to aggregate and are most abundant in the third oxygen layer, primarily due to extensive geometric relaxation-an aspect previously overlooked. This behavior is notably dependent on the concentration of Vo. This work introduces a novel theoretical framework for unraveling the complex vacancy structures in metal oxides, with potential applications in redox and catalytic chemistry.

cond-mat.mtrl-sci↗

arXiv2023

PaLM 2 Technical Report

We introduce PaLM 2, a new state-of-the-art language model that has better multilingual and reasoning capabilities and is more compute-efficient than its predecessor PaLM. PaLM 2 is a Transformer-based model trained using a mixture of objectives. Through extensive evaluations on English and multilingual language, and reasoning tasks, we demonstrate that PaLM 2 has significantly improved quality on downstream tasks across different model sizes, while simultaneously exhibiting faster and more efficient inference compared to PaLM. This improved efficiency enables broader deployment while also allowing the model to respond faster, for a more natural pace of interaction. PaLM 2 demonstrates robust reasoning capabilities exemplified by large improvements over PaLM on BIG-Bench and other reasoning tasks. PaLM 2 exhibits stable performance on a suite of responsible AI evaluations, and enables inference-time control over toxicity without additional overhead or impact on other capabilities. Overall, PaLM 2 achieves state-of-the-art performance across a diverse set of tasks and capabilities. When discussing the PaLM 2 family, it is important to distinguish between pre-trained models (of various sizes), fine-tuned variants of these models, and the user-facing products that use these models. In particular, user-facing products typically include additional pre- and post-processing steps. Additionally, the underlying models may evolve over time. Therefore, one should not expect the performance of user-facing products to exactly match the results reported in this report.

cs.CL↗

arXiv2022

Towards Understanding the Overfitting Phenomenon of Deep Click-Through Rate Prediction Models

Deep learning techniques have been applied widely in industrial recommendation systems. However, far less attention has been paid to the overfitting problem of models in recommendation systems, which, on the contrary, is recognized as a critical issue for deep neural networks. In the context of Click-Through Rate (CTR) prediction, we observe an interesting one-epoch overfitting problem: the model performance exhibits a dramatic degradation at the beginning of the second epoch. Such a phenomenon has been witnessed widely in real-world applications of CTR models. Thereby, the best performance is usually achieved by training with only one epoch. To understand the underlying factors behind the one-epoch phenomenon, we conduct extensive experiments on the production data set collected from the display advertising system of Alibaba. The results show that the model structure, the optimization algorithm with a fast convergence rate, and the feature sparsity are closely related to the one-epoch phenomenon. We also provide a likely hypothesis for explaining such a phenomenon and conduct a set of proof-of-concept experiments. We hope this work can shed light on future research on training more epochs for better performance.

cs.IR↗

arXiv2022

KEEP: An Industrial Pre-Training Framework for Online Recommendation via Knowledge Extraction and Plugging

An industrial recommender system generally presents a hybrid list that contains results from multiple subsystems. In practice, each subsystem is optimized with its own feedback data to avoid the disturbance among different subsystems. However, we argue that such data usage may lead to sub-optimal online performance because of the \textit{data sparsity}. To alleviate this issue, we propose to extract knowledge from the \textit{super-domain} that contains web-scale and long-time impression data, and further assist the online recommendation task (downstream task). To this end, we propose a novel industrial \textbf{K}nowl\textbf{E}dge \textbf{E}xtraction and \textbf{P}lugging (\textbf{KEEP}) framework, which is a two-stage framework that consists of 1) a supervised pre-training knowledge extraction module on super-domain, and 2) a plug-in network that incorporates the extracted knowledge into the downstream model. This makes it friendly for incremental training of online recommendation. Moreover, we design an efficient empirical approach for KEEP and introduce our hands-on experience during the implementation of KEEP in a large-scale industrial system. Experiments conducted on two real-world datasets demonstrate that KEEP can achieve promising results. It is notable that KEEP has also been deployed on the display advertising system in Alibaba, bringing a lift of $+5.4\%$ CTR and $+4.7\%$ RPM.

cs.IR↗

arXiv2021

CAN: Feature Co-Action for Click-Through Rate Prediction

Feature interaction has been recognized as an important problem in machine learning, which is also very essential for click-through rate (CTR) prediction tasks. In recent years, Deep Neural Networks (DNNs) can automatically learn implicit nonlinear interactions from original sparse features, and therefore have been widely used in industrial CTR prediction tasks. However, the implicit feature interactions learned in DNNs cannot fully retain the complete representation capacity of the original and empirical feature interactions (e.g., cartesian product) without loss. For example, a simple attempt to learn the combination of feature A and feature B as the explicit cartesian product representation of new features can outperform previous implicit feature interaction models including factorization machine (FM)-based models and their variations. In this paper, we propose a Co-Action Network (CAN) to approximate the explicit pairwise feature interactions without introducing too many additional parameters. More specifically, giving feature A and its associated feature B, their feature interaction is modeled by learning two sets of parameters: 1) the embedding of feature A, and 2) a Multi-Layer Perceptron (MLP) to represent feature B. The approximated feature interaction can be obtained by passing the embedding of feature A through the MLP network of feature B. We refer to such pairwise feature interaction as feature co-action, and such a Co-Action Network unit can provide a very powerful capacity to fitting complex feature interactions. Experimental results on public and industrial datasets show that CAN outperforms state-of-the-art CTR models and the cartesian product method. Moreover, CAN has been deployed in the display advertisement system in Alibaba, obtaining 12\% improvement on CTR and 8\% on Revenue Per Mille (RPM), which is a great improvement to the business.

cs.IR↗

arXiv2021

Hot Carrier Degradation in MOSFETs at Cryogenic Temperatures Down to 4.2 K

Wide attention has been focused on cryogenic CMOS (Cryo-CMOS) operation because of its wide application and the improvement of CMOS performance. However, hot carrier degradation (HCD) becomes worsening at cryogenic temperature, which affects the reliability of Cryo-CMOS. Therefore, this article investigates HCD in 0.18 um bulk CMOS at cryogenic temperature down to 4.2 K. Particularly, the relationship between HCD and the current overshoot phenomenon and the influence of substrate bias on HCD are discussed. Besides, we predict the lifetime of the device at 77 K and 4.2 K. It is concluded that cryogenic NMOS cannot reach the ten years' commercial standard lifetime at standard drain voltage (VDD). And it is predicted that the reliability requirements can be reached when VDD<1.768V/1.734V at 77K/4.2K. Differently, the lifetime of PMOS is long enough even at low temperatures.

physics.app-ph↗

arXiv2021

Cryogenic Modeling of MOSFET Device Based on BSIM and EKV Models

Kink effect is a large obstacle for the cryogenic model of inversion-type bulk silicon MOSFET devices. This letter used two methods to correct the kink effect: the modified evolutionary strategy (MES) and dual-model modeling (BSIM3v3 and EKV2.6). Both methods are based on the principle of kink effect. The first method considers impact ionization and substrate current induced body effect (SCBE), and the other considers the change of the freeze-out substrate potential. By applying the above two methods, kink can be corrected to improve the agreement between simulation data and measurement data, and obtain more accurate model parameters. These two methods can be used in further work for cryogenic device modeling and circuit design.

physics.app-ph↗

arXiv2021

Characterization and Modeling of Native MOSFETs Down to 4.2 K

The extremely low threshold voltage (Vth) of native MOSFETs (Vth~0V@300K) is conducive to the design of cryogenic circuits. Previous research on cryogenic MOSFETs mainly focused on the standard threshold voltage (SVT) and low threshold voltage (LVT) MOSFETs. In this paper, we characterize native MOSFETs within the temperature range from 300K to 4.2K. The cryogenic Vth increases up to ~0.25V (W/L=10um/10um) and the improved subthreshold swing (SS)~14.30mV/dec@4.2K. The off-state current (Ioff) and the gate-induced drain leakage (GIDL) effect are ameliorated greatly. The step-up effect caused by the substrate charge and the transconductance peak effect caused by the energy quantization in different sub-bands are also discussed. Based on the EKV model, we modified the mobility calculation equations and proposed a compact model of large size native MOSFETs suitable for the range of 300K to 4.2K. The mobility-related parameters are extracted via a machine learning approach and the temperature dependences of the scattering mechanisms are analyzed. This work is beneficial to both the research on cryogenic MOSFETs modeling and the design of cryogenic CMOS circuits for quantum chips.

physics.app-ph↗

arXiv2020

Search-based User Interest Modeling with Lifelong Sequential Behavior Data for Click-Through Rate Prediction

Rich user behavior data has been proven to be of great value for click-through rate prediction tasks, especially in industrial applications such as recommender systems and online advertising. Both industry and academy have paid much attention to this topic and propose different approaches to modeling with long sequential user behavior data. Among them, memory network based model MIMN proposed by Alibaba, achieves SOTA with the co-design of both learning algorithm and serving system. MIMN is the first industrial solution that can model sequential user behavior data with length scaling up to 1000. However, MIMN fails to precisely capture user interests given a specific candidate item when the length of user behavior sequence increases further, say, by 10 times or more. This challenge exists widely in previously proposed approaches. In this paper, we tackle this problem by designing a new modeling paradigm, which we name as Search-based Interest Model (SIM). SIM extracts user interests with two cascaded search units: (i) General Search Unit acts as a general search from the raw and arbitrary long sequential behavior data, with query information from candidate item, and gets a Sub user Behavior Sequence which is relevant to candidate item; (ii) Exact Search Unit models the precise relationship between candidate item and SBS. This cascaded search paradigm enables SIM with a better ability to model lifelong sequential behavior data in both scalability and accuracy. Apart from the learning algorithm, we also introduce our hands-on experience on how to implement SIM in large scale industrial systems. Since 2019, SIM has been deployed in the display advertising system in Alibaba, bringing 7.1\% CTR and 4.4\% RPM lift, which is significant to the business. Serving the main traffic in our real system now, SIM models user behavior data with maximum length reaching up to 54000, pushing SOTA to 54x.

cs.IR↗

arXiv2019

SIMEX Estimation in Parametric Modal Regression with Measurement Error

For a class of parametric modal regression models with measurement error, a simulation extrapolation estimation procedure is proposed in this paper for estimating the modal regression coefficients. Large sample properties of the proposed estimation procedure, including the consistency and asymptotic normality, are thoroughly investigated. Simulation studies are conducted to evaluate its robustness to potential outliers and the effectiveness in reducing the bias caused by the measurement error.

stat.ME↗

arXiv2009

Universal $\sqrt{2}\times\sqrt{2}$ structure and short-range charge order at the surfaces of BaFe$_{2-x}$Co$_{x}$As$_{2}$ compounds with various Co doping levels

The structure and electronic order at the cleaved (001) surfaces of the newly-discovered pnictide superconductors BaFe$_{2-x}$Co$_{x}$As$_{2}$ with x ranging from 0 to 0.32 are systematically investigated by scanning tunneling microscopy. A $\sqrt{2}\times\sqrt{2}$ surface structure is revealed for all the compounds, and is identified to be Ba layer with half Ba atoms lifted-off by combination with theoretical simulation. A universal short-range charge order is observed at this $\sqrt{2}\times\sqrt{2}$ surface associated with an energy gap of about 30 meV for all the compounds.

cond-mat.supr-con↗