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.
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
1 - 15 of 10069 publications
Characterizing a Memory Allocator at Warehouse Scale
Zhuangzhuang Zhou
Nilay Vaish
Patrick Xia
Christina Delimitrou
Proceedings of the 29th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, Volume 3, Association for Computing Machinery, La Jolla, CA, USA (2024), 192–206
Preview abstract
Memory allocation constitutes a substantial component of warehouse-scale computation. Optimizing the memory allocator not only reduces the datacenter tax, but also improves application performance, leading to significant cost savings.
We present the first comprehensive characterization study of TCMalloc, a warehouse-scale memory allocator used in our production fleet. Our characterization reveals a profound diversity in the memory allocation patterns, allocated object sizes and lifetimes, for large-scale datacenter workloads, as well as in their performance on heterogeneous hardware platforms. Based on these insights, we redesign TCMalloc for warehouse-scale environments. Specifically, we propose optimizations for each level of its cache hierarchy that include usage-based dynamic sizing of allocator caches, leveraging hardware topology to mitigate inter-core communication overhead, and improving allocation packing algorithms based on statistical data. We evaluate these design choices using benchmarks and fleet-wide A/B experiments in our production fleet, resulting in a 1.4% improvement in throughput and a 3.4% reduction in RAM usage for the entire fleet. At our scale, even a single percent CPU or memory improvement translates to significant savings in server costs.
View details
Preview abstract
Middle-mile logistics describes the problem of routing shipments through a network of hubs while respecting deadlines upon arrival. We consider that the hubs are linked by predefined lines, to which we have to assign vehicles. A very challenging aspect of the problem comes from the finite capacity of the vehicles: allocating a shipment to a given vehicle might block another one from using the same vehicle.
Typical exact solution methods, based on a multicommodity-flow formulation, scale poorly with the problem size and real-world instances become quickly intractable. Instead, we turn to reinforcement learning (RL) by rephrasing the middle-mile problem as a multi-objective Markov decision process, where the state is a graph: the lines (edges) between the hubs and the parcels (nodes). At each round, we assign one shipment to a vehicle or decide that it stays in the same hub. The key ingredients of our proposed method are the extraction of small feature graphs from the state and the combination of graph neural networks (GNN) with model-free RL.
We use the PPO (proximal policy optimization) algorithm, which maintains both an actor and a critic, while being able to cope with a varying number of actions depending on the state. We compare linear functions and GraphNet (a particular kind of GNN) to approximate the policy and value functions. GNNs can deliver up to 40% more shipments than a linear function and both approaches scale well with the number of shipments per truck.
View details
Analyzing Prospects for Quantum Advantage in Topological Data Analysis
Dominic W. Berry
Yuan Su
Casper Gyurik
Robbie King
Joao Basso
Abhishek Rajput
Nathan Wiebe
Vedran Djunko
PRX Quantum, 5 (2024), pp. 010319
Preview abstract
Lloyd et al. were first to demonstrate the promise of quantum algorithms for computing Betti numbers in persistent homology (a way of characterizing topological features of data sets). Here, we propose, analyze, and optimize an improved quantum algorithm for topological data analysis (TDA) with reduced scaling, including a method for preparing Dicke states based on inequality testing, a more efficient amplitude estimation algorithm using Kaiser windows, and an optimal implementation of eigenvalue projectors based on Chebyshev polynomials. We compile our approach to a fault-tolerant gate set and estimate constant factors in the Toffoli complexity. Our analysis reveals that super-quadratic quantum speedups are only possible for this problem when targeting a multiplicative error approximation and the Betti number grows asymptotically. Further, we propose a dequantization of the quantum TDA algorithm that shows that having exponentially large dimension and Betti number are necessary, but insufficient conditions, for super-polynomial advantage. We then introduce and analyze specific problem examples for which super-polynomial advantages may be achieved, and argue that quantum circuits with tens of billions of Toffoli gates can solve some seemingly classically intractable instances.
View details
Bridging the Gap: Unpacking the Hidden Challenges in Knowledge Distillation for Online Ranking Systems
Shuo Yang
Aniruddh Nath
Yang Liu
Li Wei
Shawn Andrews
Maciej Kula
Jarrod Kahn
Zhe Zhao
Lichan Hong
Preview abstract
Knowledge Distillation (KD) is a powerful approach for compressing large models into smaller, more efficient models, particularly beneficial for latency-sensitive applications like recommender systems. However, current KD research predominantly focuses on Computer Vision (CV) and NLP tasks, overlooking unique data characteristics and challenges inherent to recommender systems. This paper addresses these overlooked challenges, specifically: (1) mitigating data distribution shifts between teacher and student models, (2) efficiently identifying optimal teacher configurations within time and budgetary constraints, and (3) enabling computationally efficient and rapid sharing of teacher labels to support multiple students. We present a robust KD system developed and rigorously evaluated on multiple large-scale personalized video recommendation systems within Google. Our live experiment results demonstrate significant improvements in student model performance while ensuring the consistent and reliable generation of high-quality teacher labels from continuous data streams.
View details
Preview abstract
We introduce SynCLR, a novel approach for learning visual representations exclusively from synthetic images and synthetic captions, without any real data. We synthesize a large dataset of image captions using LLMs, then use an off-the-shelf text-to-image model to generate multiple images corresponding to each synthetic caption. We perform visual representation learning on these synthetic images via contrastive learning, treating images sharing the same caption as positive pairs. The resulting representations transfer well to many downstream tasks, competing favorably with other general-purpose visual representation learners such as CLIP and DINO v2 in image classification tasks. Furthermore, in dense prediction tasks such as semantic segmentation, SynCLR outperforms previous self-supervised methods by a significant margin, e.g., improving over MAE and iBOT by 6.2 and 4.3 mIoU on ADE20k for ViT-B/16.
View details
Optimizing quantum gates towards the scale of logical qubits
Alexandre Bourassa
Andrew Dunsworth
Will Livingston
Vlad Sivak
Trond Andersen
Yaxing Zhang
Desmond Chik
Jimmy Chen
Charles Neill
Alejo Grajales Dau
Anthony Megrant
Alexander Korotkov
Vadim Smelyanskiy
Yu Chen
Nature Communications, 15 (2024), pp. 2442
Preview abstract
A foundational assumption of quantum error correction theory is that quantum gates can be scaled to large processors without exceeding the error-threshold for fault tolerance. Two major challenges that could become fundamental roadblocks are manufacturing high-performance quantum hardware and engineering a control system that can reach its performance limits. The control challenge of scaling quantum gates from small to large processors without degrading performance often maps to non-convex, high-constraint, and time-dynamic control optimization over an exponentially expanding configuration space. Here we report on a control optimization strategy that can scalably overcome the complexity of such problems. We demonstrate it by choreographing the frequency trajectories of 68 frequency-tunable superconducting qubits to execute single- and two-qubit gates while mitigating computational errors. When combined with a comprehensive model of physical errors across our processor, the strategy suppresses physical error rates by ~3.7× compared with the case of no optimization. Furthermore, it is projected to achieve a similar performance advantage on a distance-23 surface code logical qubit with 1057 physical qubits. Our control optimization strategy solves a generic scaling challenge in a way that can be adapted to a variety of quantum operations, algorithms, and computing architectures.
View details
Preview abstract
Vortex is an exabyte scale structured storage system built for streaming and batch analytics. It supports high-throughput batch and stream ingestion. For the user, it supports both batch-oriented and stream-based processing on the ingested data.
View details
V2Meow: Meowing to the Visual Beat via Video-to-Music Generation
Chris Donahue
Dima Kuzmin
Judith Li
Kun Su
Mauro Verzetti
Qingqing Huang
Yu Wang
Vol. 38 No. 5: AAAI-24 Technical Tracks 5, AAAI Press (2024), pp. 4952-4960
Preview abstract
Video-to-music generation demands both a temporally localized high-quality listening experience and globally aligned video-acoustic signatures. While recent music generation models excel at the former through advanced audio codecs, the exploration of video-acoustic signatures has been confined to specific visual scenarios. In contrast, our research confronts the challenge of learning globally aligned signatures between video and music directly from paired music and videos, without explicitly modeling domain-specific rhythmic or semantic relationships. We propose V2Meow, a video-to-music generation system capable of producing high-quality music audio for a diverse range of video input types using a multi-stage autoregressive model. Trained on 5k hours of music audio clips paired with video frames mined from in-the-wild music videos, V2Meow is competitive with previous domain-specific models when evaluated in a zero-shot manner. It synthesizes high-fidelity music audio waveforms solely by conditioning on pre-trained general purpose visual features extracted from video frames, with optional style control via text prompts. Through both qualitative and quantitative evaluations, we demonstrate that our model outperforms various existing music generation systems in terms of visual-audio correspondence and audio quality. Music samples are available at tinyurl.com/v2meow.
View details
Preview abstract
Reinforcement can be a useful tool to solve combinatorial problems, even in the presence of constraints. This presentation details two use cases: one industrial application in the field of logistics, one of a more abstract problem in combinatorial optimization.
View details
Preview abstract
Floods are one of the most common natural disasters, with a disproportionate impact in developing countries that often lack dense streamflow gauge networks. Accurate and timely warnings are critical for mitigating flood risks, but hydrological simulation models typically must be calibrated to long data records in each watershed. Here we show that AI-based forecasting achieves reliability in predicting extreme riverine events in ungauged watersheds at up to a 5-day lead time that is similar to or better than the reliability of nowcasts (0-day lead time) from a current state of the art global modeling system (the Copernicus Emergency Management Service Global Flood Awareness System). Additionally, we achieve accuracies over 5-year return period events that are similar to or better than current accuracies over 1-year return period events. This means that AI can provide flood warnings earlier and over larger and more impactful events in ungauged basins. The model developed in this paper was incorporated into an operational early warning system that produces publicly available (free and open) forecasts in real time in over 80 countries. This work highlights a need for increasing the availability of hydrological data to continue to improve global access to reliable flood warnings.
View details
Learning to Rewrite Prompts for Personalized Text Generation
Qiaozhu Mei
Proceedings of the ACM Web Conference 2024
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
Preview abstract
While large, generative, multilingual models are rapidly being developed and deployed, their safety and fairness evaluations primarily hinge on resources collected in the English language and some limited translations. This has been demonstrated to be insufficient, and severely lacking in nuances of unsafe language and stereotypes prevalent in different languages and the geographical pockets they are prevalent in. Gathering these resources, at scale, in varied languages and regions also poses a challenge as it requires expansive sociolinguistic knowledge and can also be prohibitively expensive. We utilize an established methodology of coupling LLM generations with distributed annotations to overcome these gaps and create the resource SeeGULL Multilingual, spanning 20 languages across 23 regions.
View details
Preview abstract
Specialized Large multi-modal models (LMMs) have exhibited remarkable performance across numerous tasks, however, generalist LMMs suffer from performance degradation when training with a large collection of tasks. Recent research suggests Mixture of Experts (MoE) Models help instruction tuning, however, for LMMs of parameter size around O(50-100B), the prohibitive cost of replicating and storing the expert models severely limits the number of experts we can use.
We propose Omni-SMoLA that softly mixes many multimodal low rank experts to large models without introducing significant new parameter count compared to conventional MoE models. The core idea is that the large model provides a foundational backbone and different lightweight experts learn specialized knowledge residually. Extensive experiments demonstrate that the SMoLA approach helps improve the generalist performance across a broad range of visual question answering and captioning tasks, achieving a new state-of-the-art generalist performance that matches or outperforms single specialized LMM baselines.
View details
The effect of uncertainty in humidity and model parameters on the prediction of contrail energy forcing
Marc Shapiro
Zebediah Engberg
Tharun Sankar
Marc E.J. Stettler
Roger Teoh
Ulrich Schumann
Susanne Rohs
Erica Brand
Environmental Research Communications, 6 (2024), pp. 095015
Preview abstract
Previous work has shown that while the net effect of aircraft condensation trails (contrails) on the climate is warming, the exact magnitude of the energy forcing per meter of contrail remains uncertain. In this paper, we explore the skill of a Lagrangian contrail model (CoCiP) in identifying flight segments with high contrail energy forcing. We find that skill is greater than climatological predictions alone, even accounting for uncertainty in weather fields and model parameters. We estimate the uncertainty due to humidity by using the ensemble ERA5 weather reanalysis from the European Centre for Medium-Range Weather Forecasts (ECMWF) as Monte Carlo inputs to CoCiP. We unbias and correct under-dispersion on the ERA5 humidity data by forcing a match to the distribution of in situ humidity measurements taken at cruising altitude. We take CoCiP energy forcing estimates calculated using one of the ensemble members as a proxy for ground truth, and report the skill of CoCiP in identifying segments with large positive proxy energy forcing. We further estimate the uncertainty due to model parameters in CoCiP by performing Monte Carlo simulations with CoCiP model parameters drawn from uncertainty distributions consistent with the literature. When CoCiP outputs are averaged over seasons to form climatological predictions, the skill in predicting the proxy is 44%, while the skill of per-flight CoCiP outputs is 84%. If these results carry over to the true (unknown) contrail EF, they indicate that per-flight energy forcing predictions can reduce the number of potential contrail avoidance route adjustments by 2x, hence reducing both the cost and fuel impact of contrail avoidance.
View details
Preview abstract
Current approaches to Analog Layout Automation
apply ML techniques such as Graph Convolutional Neural
Networks (GCN) to translate netlist to layout. While these ML
approaches have proven to be effective, they lack the powerful
reasoning capabilities, an intuitive human interface, and standard
evaluation benchmarks that have been improving at a rapid de-
velopment pace in Large Language Models (LLMs). The GLayout
framework introduced in this work translates analog layout into
an expressive, technology generic, compact text representation.
Then, an LLM is taught to understand analog layout through
fine-tuning and in-context learning using Retrieval Augmented
Generation (RAG). The LLM is able to successfully layout unseen
circuits based on new information provided in-context. We train
3.8, 7, and 22 Billion parameter quantized LLMs on a dataset
of less than 50 unique circuits, and text documents providing
layout knowledge. The 22B parameter model is tuned in 2 hours
on a single NVIDIA A100 GPU. The open-source evaluation
set is proposed as an automation benchmark for LLM layout
automation tasks, and ranges from 2-transistor circuits to a
∆Σ ADC. The 22B model completes 70% of the tasks in the
evaluation set, and is able to pass DRC and LVS verification on
unseen 4 transistor blocks.
View details
