Publications

Our teams aspire to make discoveries that impact everyone, and core to our approach is sharing our research and tools to fuel progress in the field.

people standing in front of a screen with images and a chipboard

Our teams aspire to make discoveries that impact everyone, and core to our approach is sharing our research and tools to fuel progress in the field.

Sort By
  • Title
  • Title, descending
  • Year
  • Year, descending
1 - 15 of 10100 publications
    Preview abstract At Google, we’ve been running a quarterly large-scale survey with developers since 2018. In this article, we will discuss how we run EngSat, some of our key learnings over the past 6 years, and how we’ve evolved our approach to meet new needs and challenges. View details
    Visual Program Tuning: Training Large Multimodal Models to Reason like Programs
    Yushi Hu
    Krishna Viswanathan
    Kenji Hata
    Enming Luo
    Ranjay Krishna
    Ariel Fuxman
    Conference on Computer Vision and Pattern Recognition (2024)
    Preview abstract Solving complex visual tasks (e.g., “Who invented the musical instrument on the right?”) involves back-and-forth between visual processing and reasoning. Visual programming is a recent multimodal framework that has shown promise in conducting visual reasoning in an interpretable and compositional manner. However, this framework is error-prone—it can lead to a wrong answer whenever the program itself is wrong, or when any of the steps of the program are solved incorrectly, thus leading to worse overall performance than end-to-end systems trained with labeled data. Moreover, it is inefficient to involve multiple steps (i.e., generating and then running programs) during inference. Ideally, a single large multimodal model (LMM) should directly conduct similar reasoning and yield the correct answer. In this work, we propose Visual Program Tuning (VPT), which leverages visual programs for teaching LLMs to reason via instruction tuning. VPT rewrites the execution traces of visual programs as chain-of-thought reasoning steps, and tunes an LMM to output not only the label but its reasoning as well. Extensive experiments on complex vision tasks show that models trained with VPT achieve state-of-the-art accuracy while being able to produce interpretable and faithful reasoning steps. PaLI-X + VPT outperforms all existing LMMs on a wide range of visual tasks, improving performance on counting, spatial relations, and compositional reasoning tasks. VPT is also helpful for quick adaptation on new tasks. Our experiments on content moderation show that fine-tuning LMMs with program-augmented examples is more sample efficient than traditional supervised training. View details
    Scaling Up LLM Reviews for Google Ads Content Moderation
    Ariel Fuxman
    Chih-Chun Chia
    Dongjin Kwon
    Enming Luo
    Mehmet Tek
    Ranjay Krishna
    Tiantian Fang
    Tushar Dogra
    Yu-Han Lyu
    (2024)
    Preview abstract Large language models (LLMs) are powerful tools for content moderation but LLM inference costs and latency on large volumes of data, such as the Google Ads repository, are prohibitive for their casual usage. This study is focused on scaling up LLM reviews for content moderation in Google Ads. First, we use heuristics to select candidates via filtering and duplicate removal, and create clusters of ads for which we select one representative ad per cluster. Then, LLMs are used to review only the representative ads. Finally we propagate the LLM decisions for representative ads back to their clusters. This method reduces the number of reviews by more than 3 orders of magnitude while achieving a 2x recall compared to a non-LLM model as a baseline. Note that, the success of this approach is a strong function of the representations used in clustering and label propagation; we observed that cross-modal similarity representations yield better results than uni-modal representations. View details
    Preview abstract Effective model calibration is a critical and indispensable component in developing Media Mix Models (MMMs). One advantage of Bayesian-based MMMs lies in their capacity to accommodate the information from experiment results and the modelers' domain knowledge about the ad effectiveness by setting priors for the model parameters. However, it remains ambiguous about how and which Bayesian priors should be tuned for calibration purpose. In this paper, we propose a new calibration method through model reparameterization. The reparameterized model includes Return on Ads Spend (ROAS) as a model parameter, enabling straightforward adjustment of its prior distribution to align with either experiment results or the modeler's prior knowledge. The proposed method also helps address several key challenges regarding combining MMMs and incrementality experiments. We use simulations to demonstrate that our approach can significantly reduce the bias and uncertainty in the resultant posterior ROAS estimates. View details
    Preview abstract This is an invited OFC 2024 conference workshop talk regarding a new type of lower-power datacenter optics design choice: linear pluggable optics. In this talk I will discuss the fundamental performance constraints facing linear pluggable optics and their implications on DCN and ML use cases View details
    Modeling Recommender Ecosystems: Research Challenges at the Intersection of Mechanism Design, Reinforcement Learning and Generative Models
    Martin Mladenov
    Proceedings of the 38th Annual AAAI Conference on Artificial Intelligence (AAAI-24), Vancouver (2024) (to appear)
    Preview abstract Modern recommender systems lie at the heart of complex ecosystems that couple the behavior of users, content providers, advertisers, and other actors. Despite this, the focus of the majority of recommender research---and most practical recommenders of any import---is on the \emph{local, myopic} optimization of the recommendations made to individual users. This comes at a significant cost to the \emph{long-term utility} that recommenders could generate for its users. We argue that explicitly modeling the incentives and behaviors of all actors in the system---and the interactions among them induced by the recommender's policy---is strictly necessary if one is to maximize the value the system brings to these actors and improve overall ecosystem ``health.'' Doing so requires: optimization over long horizons using techniques such as \emph{reinforcement learning}; making inevitable tradeoffs among the utility that can be generated for different actors using the methods of \emph{social choice}; reducing information asymmetry, while accounting for incentives and strategic behavior, using the tools of \emph{mechanism design}; better modeling of both user and item-provider behaviors by incorporating notions from \emph{behavioral economics and psychology}; and exploiting recent advances in \emph{generative and foundation models} to make these mechanisms interpretable and actionable. We propose a conceptual framework that encompasses these elements, and articulate a number of research challenges that emerge at the intersection of these different disciplines. View details
    On the Benefits of Traffic “Reprofiling” The Multiple Hops Case – Part I
    Henry Sariowan
    Jiaming Qiu
    Jiayi Song
    Roch Guerin
    IEEE/ACM Transactions on Networking (2024)
    Preview abstract Abstract—This paper considers networks where user traffic is regulated through deterministic traffic profiles, e.g. token buckets, and requirescleanguaranteed hard delay bounds. The network’s goal is to minimize the resources it needs to meet those cleanrequirementsbounds. The paper explores how reprofiling, i.e. proactively modifying how user traffic enters the network, can be of benefit. Reprofiling produces “smoother” flows but introduces an up-front access delay that forces tighter network delays. The paper explores this trade-off and demonstrates that, unlike what holds in the single-hop case, reprofiling can be of benefit even when “optimal”cleansophisticated schedulers are available at each hop. View details
    Preview abstract We present Spectron, a novel approach to adapting pre-trained large language models (LLMs) to perform spoken question answering (QA) and speech continuation. By endowing the LLM with a pre-trained speech encoder, our model becomes able to take speech inputs and generate speech outputs. The entire system is trained endto-end and operates directly on spectrograms, simplifying our architecture. Key to our approach is a training objective that jointly supervises speech recognition, text continuation, and speech synthesis using only paired speech-text pairs, enabling a ‘cross-modal’ chain-of-thought within a single decoding pass. Our method surpasses existing spoken language models in speaker preservation and semantic coherence. Furthermore, the proposed model improves upon direct initialization in retaining the knowledge of the original LLM as demonstrated through spoken QA datasets. We release our audio samples and spoken QA dataset via our website. View details
    Preview abstract We study the price of anarchy of the generalized second-price auction where bidders are value maximizers (i.e., autobidders). We show that in general the price of anarchy can be as bad as 0. For comparison, the price of anarchy of running VCG is 1/2 in the autobidding world. We further show a fined-grained price of anarchy with respect to the discount factors (i.e., the ratios of click probabilities between lower slots and the highest slot in each auction) in the generalized second-price auction, which highlights the qualitative relation between the smoothness of the discount factors and the efficiency of the generalized second-price auction. View details
    Preview abstract Recent developments in large language models (LLMs) have shown promise in their ability to generate synthetic query-document pairs by prompting LLMs with as few as 8 demonstrations \cite{dai2022promptagator}. This has enabled building better IR models especially for tasks which have no training data readily available. Typically, such synthetic query generation (QGen) approaches condition on an input context (e.g. document) and generate a query that is relevant to that context or condition the QGen model additionally on the relevance label (e.g. relevant vs irrelevant) to generate queries across relevance buckets. However, we find that such QGen approaches are sub-optimal as it requires the model to reason about the desired label and the input from only a handful of examples, which is not trivial, especially when the relevance buckets are nuanced. In this work, we propose to reduce this burden of LLMs by generating queries simultaneously for different labels (e.g. relevance buckets). We hypothesize that instead of asking the model to generate, say, an irrelevant query given an input context, asking the model to generate an irrelevant query with respect to a relevant query is a much simpler task setup for the model to reason about. Extensive experimentation across seven IR datasets shows that synthetic queries generated in such a fashion translates to a better downstream performance, suggesting that the generated queries are indeed of higher quality. View details
    General Identifiability and Achievability for Causal Representation Learning
    Burak Varici
    Emre Acarturk
    Ali Tajer
    AISTATS 2024 (Oral), Oral Talk at NeurIPS Causal Representation Learning Workshop 2023. (2024)
    Preview abstract This paper focuses on causal representation learning (CRL) under a general nonparametric latent causal model and a general transformation model that maps the latent data to the observational data. It establishes identifiability and achievability results using two hard uncoupled interventions per node in the latent causal graph. Notably, one does not know which pair of intervention environments have the same node intervened (hence, uncoupled). For identifiability, the paper establishes that perfect recovery of the latent causal model and variables is guaranteed under uncoupled interventions. For achievability, an algorithm is designed that uses observational and interventional data and recovers the latent causal model and variables with provable guarantees. This algorithm leverages score variations across different environments to estimate the inverse of the transformer and, subsequently, the latent variables. The analysis, additionally, recovers the identifiability result for two hard coupled interventions, that is when metadata about the pair of environments that have the same node intervened is known. This paper also shows that when observational data is available, additional faithfulness assumptions that are adopted by the existing literature are unnecessary View details
    LLM Comparator: Visual Analytics for Side-by-Side Evaluation of Large Language Models
    Michael Xieyang Liu
    Krystal Kallarackal
    Extended Abstracts of the CHI Conference on Human Factors in Computing Systems (CHI EA '24), ACM (2024)
    Preview abstract Automatic side-by-side evaluation has emerged as a promising approach to evaluating the quality of responses from large language models (LLMs). However, analyzing the results from this evaluation approach raises scalability and interpretability challenges. In this paper, we present LLM Comparator, a novel visual analytics tool for interactively analyzing results from automatic side-by-side evaluation. The tool supports interactive workflows for users to understand when and why a model performs better or worse than a baseline model, and how the responses from two models are qualitatively different. We iteratively designed and developed the tool by closely working with researchers and engineers at Google. This paper details the user challenges we identified, the design and development of the tool, and an observational study with participants who regularly evaluate their models. View details
    Preview abstract Facilitated by large language models (LLMs), personalized text generation has become a rapidly growing research direction. Most existing studies focus on designing specialized models for a particular domain, or they require fine-tuning the LLMs to generate personalized text. We consider a typical scenario in which the large language model, which generates personalized output, is frozen and can only be accessed through APIs. Under this constraint, all one can do is to improve the input text (i.e., text prompts) sent to the LLM, a procedure that is usually done manually. In this paper, we propose a novel method to automatically revise prompts for personalized text generation. The proposed method takes the initial prompts generated by a state-of-the-art, multistage framework for personalized generation and rewrites a few critical components that summarize and synthesize the personal context. The prompt rewriter employs a training paradigm that chains together supervised learning (SL) and reinforcement learning (RL), where SL reduces the search space of RL and RL facilitates end-to-end training of the rewriter. Using datasets from three representative domains, we demonstrate that the rewritten prompts outperform both the original prompts and the prompts optimized via supervised learning or reinforcement learning alone. In-depth analysis of the rewritten prompts shows that they are not only human readable, but also able to guide manual revision of prompts when there is limited resource to employ reinforcement learning to train the prompt rewriter, or when it is costly to deploy an automatic prompt rewriter for inference. View details
    USER-LLM: Efficient LLM Contextualization with User Embedding
    Jiaxing Wu
    Neo Wu
    Devora Berlowitz
    Sushant Prakash
    Bradley Green
    Shawn O'Banion
    Jun Xie
    ArXiv (2024) (to appear)
    Preview abstract Large language models (LLMs) have revolutionized natural language processing. However, effectively incorporating complex and potentially noisy user interaction data remains a challenge. To address this, we propose User-LLM, a novel framework that leverages user embeddings to contextualize LLMs. These embeddings, distilled from diverse user interactions using self-supervised pretraining, capture latent user preferences and their evolution over time. We integrate these user embeddings with LLMs through cross-attention and soft-prompting, enabling LLMs to dynamically adapt to user context. Our comprehensive experiments on MovieLens, Amazon Review, and Google Local Review datasets demonstrate significant performance gains across various tasks. Notably, our approach outperforms text-prompt-based contextualization on long sequence tasks and tasks that require deep user understanding while being computationally efficient. We further incorporate Perceiver layers to streamline the integration between user encoders and LLMs, reducing computational demands. View details
    Preview abstract This paper reflects on work at Google over the past decade to address common types of software safety and security defects. Our experience has shown that software safety is an emergent property of the software and tooling ecosystem it is developed in and the production environment into which it is deployed. Thus, to effectively prevent common weaknesses at scale, we need to shift-left the responsibility for ensuring safety and security invariants to the end-to-end developer ecosystem, that is, programming languages, software libraries, application frameworks, build and deployment tooling, the production platform and its configuration surfaces, and so forth. Doing so is practical and cost effective when developer ecosystems are designed with application archetypes in mind, such as web or mobile apps: The design of the developer ecosystem can address threat model aspects that apply commonly to all applications of the respective archetype, and investments to ensure safety invariants at the ecosystem level amortize across many applications. Applying secure-by-design principles to developer ecosystems at Google has achieved drastic reduction and in some cases near-zero residual rates of common classes of defects, across hundreds of applications being developed by thousands of developers. View details