Showing 1–50 of 107 results for author: Gardner, J

Self-Driving Datasets: From 20 Million Papers to Nuanced Biomedical Knowledge at Scale

Authors: Haydn Jones, Yimeng Zeng, Alden Rose, Li S. Yifei, Yining Huang, Kaiwen Wu, Jiaming Liang, Maggie Ziyu Huan, Yoseph Barash, Cesar de la Fuente-Nunez, Osbert Bastani, Zachary Ives, Mark Yatskar, Jacob R. Gardner

Abstract: Manually curated biomedical repositories -- spanning bioactivity, genomics, and chemistry -- are expensive to maintain, lag behind primary literature, and discard experimental context, obscuring nuances needed to assess data correctness and coverage. We show that PubMed itself can be autonomously and cost-effectively turned into structured datasets that are larger, more nuanced, and more accurate… ▽ More Manually curated biomedical repositories -- spanning bioactivity, genomics, and chemistry -- are expensive to maintain, lag behind primary literature, and discard experimental context, obscuring nuances needed to assess data correctness and coverage. We show that PubMed itself can be autonomously and cost-effectively turned into structured datasets that are larger, more nuanced, and more accurate than the curated databases they replace. We present three coupled contributions: (1) an LLM-based entity-tagging pipeline, grounded in nine biomedical ontologies, that tags 4.5B entities across 19 categories in a 22.5M-paper, 2.5T-token PubMed corpus; (2) hybrid sparse-dense retrieval supporting entity-filtered semantic queries over the tagged corpus; and (3) Starling, a multi-agent deep research system that, given only a natural-language task description, designs precision- and recall-targeted retrieval filters, induces an extraction schema, and emits structured records with nuance-rich fields and supporting passages. Across six tasks -- blood-brain barrier permeability, oral bioavailability, acute toxicity (LD50), gene-disease associations, protein subcellular localization, and chemical reactions -- Starling produces ~6.3M records (91K-3M per task); several are, to our knowledge, the largest public datasets for their property. Frontier-model rejection of our extractions is 0.6-7.7% across tasks, far below error rates we measure on widely used curated counterparts (e.g., 16.5% on BBB_Martins, 7.3% on Bioavailability_Ma). Beyond scale and accuracy, the supporting passages carry nuance tabular databases discard -- e.g., oral bioavailability may depend on fed vs. fasted state. Together, the corpus, retrieval, and agent establish a foundation for AI-driven therapeutic design. Code and datasets: https://github.com/starling-labs/starling. △ Less

Submitted 16 June, 2026; v1 submitted 7 May, 2026; originally announced May 2026.

Adversarial Query Synthesis via Bayesian Optimization

Authors: Jeffrey Tao, Yimeng Zeng, Haydn Thomas Jones, Natalie Maus, Osbert Bastani, Jacob R. Gardner, Ryan Marcus

Abstract: Benchmark workloads are extremely important to the database management research community, especially as more machine learning components are integrated into database systems. Here, we propose a Bayesian optimization technique to automatically search for difficult benchmark queries, significantly reducing the amount of manual effort usually required. In preliminary experiments, we show that our ap… ▽ More Benchmark workloads are extremely important to the database management research community, especially as more machine learning components are integrated into database systems. Here, we propose a Bayesian optimization technique to automatically search for difficult benchmark queries, significantly reducing the amount of manual effort usually required. In preliminary experiments, we show that our approach can generate queries with more than double the optimization headroom compared to existing benchmarks. △ Less

Submitted 2 March, 2026; originally announced March 2026.

Beyond a Single Extractor: Re-thinking HTML-to-Text Extraction for LLM Pretraining

Authors: Jeffrey Li, Josh Gardner, Doug Kang, Fangping Shi, Karanjeet Singh, Chun-Liang Li, Herumb Shandilya, David Hall, Oncel Tuzel, Percy Liang, Ludwig Schmidt, Hadi Pour Ansari, Fartash Faghri

Abstract: One of the first pre-processing steps for constructing web-scale LLM pretraining datasets involves extracting text from HTML. Despite the immense diversity of web content, existing open-source datasets predominantly apply a single fixed extractor to all webpages. In this work, we investigate whether this practice leads to suboptimal coverage and utilization of Internet data. We first show that whi… ▽ More One of the first pre-processing steps for constructing web-scale LLM pretraining datasets involves extracting text from HTML. Despite the immense diversity of web content, existing open-source datasets predominantly apply a single fixed extractor to all webpages. In this work, we investigate whether this practice leads to suboptimal coverage and utilization of Internet data. We first show that while different extractors may lead to similar model performance on standard language understanding tasks, the pages surviving a fixed filtering pipeline can differ substantially. This suggests a simple intervention: by taking a Union over different extractors, we can increase the token yield of DCLM-Baseline by up to 71% while maintaining benchmark performance. We further show that for structured content such as tables and code blocks, extractor choice can significantly impact downstream task performance, with differences of up to 10 percentage points (p.p.) on WikiTQ and 3 p.p. on HumanEval. △ Less

Submitted 23 February, 2026; originally announced February 2026.

Stochastic Gradient Variational Inference with Price's Gradient Estimator from Bures-Wasserstein to Parameter Space

Authors: Kyurae Kim, Qiang Fu, Yi-An Ma, Jacob R. Gardner, Trevor Campbell

Abstract: For approximating a target distribution given only its unnormalized log-density, stochastic gradient-based variational inference (VI) algorithms are a popular approach. For example, Wasserstein VI (WVI) and black-box VI (BBVI) perform gradient descent in measure space (Bures-Wasserstein space) and parameter space, respectively. Previously, for the Gaussian variational family, convergence guarantee… ▽ More For approximating a target distribution given only its unnormalized log-density, stochastic gradient-based variational inference (VI) algorithms are a popular approach. For example, Wasserstein VI (WVI) and black-box VI (BBVI) perform gradient descent in measure space (Bures-Wasserstein space) and parameter space, respectively. Previously, for the Gaussian variational family, convergence guarantees for WVI have shown superiority over existing results for black-box VI with the reparametrization gradient, suggesting the measure space approach might provide some unique benefits. In this work, however, we close this gap by obtaining identical state-of-the-art iteration complexity guarantees for both. In particular, we identify that WVI's superiority stems from the specific gradient estimator it uses, which BBVI can also leverage with minor modifications. The estimator in question is usually associated with Price's theorem and utilizes second-order information (Hessians) of the target log-density. We will refer to this as Price's gradient. On the flip side, WVI can be made more widely applicable by using the reparametrization gradient, which requires only gradients of the log-density. We empirically demonstrate that the use of Price's gradient is the major source of performance improvement. △ Less

Submitted 19 May, 2026; v1 submitted 20 February, 2026; originally announced February 2026.

Comments: Accepted to ICML'26

MagneX: A High-Performance, GPU-Enabled, Data-Driven Micromagnetics Solver for Spintronics

Authors: Andy Nonaka, Yingheng Tang, Julian C. LePelch, Prabhat Kumar, Weiqun Zhang, Jorge A. Munoz, Christian Fernandez-Soria, Cesar Diaz, David J. Gardner, Zhi Jackie Yao

Abstract: In order to comprehensively investigate the multiphysics coupling in spintronic devices, it is essential to parallelize and utilize GPU-acceleration to address the spatial and temporal disparities inherent in the relevant physics. Additionally, the use of cutting-edge time integration libraries as well as machine learning (ML) approaches to replace and potentially accelerate expensive computationa… ▽ More In order to comprehensively investigate the multiphysics coupling in spintronic devices, it is essential to parallelize and utilize GPU-acceleration to address the spatial and temporal disparities inherent in the relevant physics. Additionally, the use of cutting-edge time integration libraries as well as machine learning (ML) approaches to replace and potentially accelerate expensive computational routines are attractive capabilities to enhance modeling capabilities moving forward. Leveraging the Exascale Computing Project software framework AMReX, as well as SUNDIALS time-integration libraries and python-based ML workflows, we have developed an open-source micromagnetics modeling tool called MagneX. This tool incorporates various crucial magnetic coupling mechanisms, including Zeeman coupling, demagnetization coupling, crystalline anisotropy interaction, exchange coupling, and Dzyaloshinskii-Moriya interaction (DMI) coupling. We demonstrate the GPU performance and scalability of the code and rigorously validate MagneX's functionality using the mumag standard problems and widely-accepted DMI benchmarks. Furthermore, we demonstrate the data-driven capability of MagneX by replacing the computationally-expensive demagnetization physics with neural network libraries trained from our simulation data. With the capacity to explore complete physical interactions, this innovative approach offers a promising pathway to better understand and develop fully integrated spintronic and electronic systems. △ Less

Submitted 13 February, 2026; v1 submitted 12 February, 2026; originally announced February 2026.

Purely Agent-Driven Black-Box Optimization for Biological Design

Authors: Natalie Maus, Yimeng Zeng, Haydn Thomas Jones, Yining Huang, Gaurav Ng Goel, Alden Rose, Kyurae Kim, Hyun-Su Lee, Marcelo Der Torossian Torres, Fangping Wan, Cesar de la Fuente-Nunez, Mark Yatskar, Osbert Bastani, Jacob R. Gardner

Abstract: Many key challenges in biological design -- such as small-molecule drug discovery, antimicrobial peptide development, and protein engineering -- can be framed as black-box optimization over vast, complex structured spaces. Existing methods rely mainly on raw structural data and struggle to exploit the rich scientific literature. While large language models (LLMs) have been added to these pipelines… ▽ More Many key challenges in biological design -- such as small-molecule drug discovery, antimicrobial peptide development, and protein engineering -- can be framed as black-box optimization over vast, complex structured spaces. Existing methods rely mainly on raw structural data and struggle to exploit the rich scientific literature. While large language models (LLMs) have been added to these pipelines, they have been confined to narrow roles within structure-centered optimizers. We instead cast biological black-box optimization as an agent-driven, language-based reasoning process. We introduce Purely Agent-driven BLack-box Optimization (PABLO), a hierarchical agentic system that uses scientific LLMs pretrained on chemistry and biology literature to generate and iteratively refine biological candidates. On both the standard GuacaMol molecular design and antimicrobial peptide optimization tasks, PABLO achieves state-of-the-art performance, substantially improving sample efficiency and final objective values over established baselines. Compared to prior optimization methods that incorporate LLMs, PABLO achieves competitive token usage per run despite relying on LLMs throughout the optimization loop. Beyond raw performance, the agentic formulation offers key advantages for realistic design: it naturally incorporates semantic task descriptions, retrieval-augmented domain knowledge, and complex constraints. In follow-up in vitro validation, PABLO-optimized peptides showed strong activity against drug-resistant pathogens, underscoring the practical potential of PABLO for therapeutic discovery. △ Less

Submitted 7 May, 2026; v1 submitted 29 January, 2026; originally announced January 2026.

Knowledge Gradient for Preference Learning

Authors: Kaiwen Wu, Jacob R. Gardner

Abstract: The knowledge gradient is a popular acquisition function in Bayesian optimization (BO) for optimizing black-box objectives with noisy function evaluations. Many practical settings, however, allow only pairwise comparison queries, yielding a preferential BO problem where direct function evaluations are unavailable. Extending the knowledge gradient to preferential BO is hindered by its computational… ▽ More The knowledge gradient is a popular acquisition function in Bayesian optimization (BO) for optimizing black-box objectives with noisy function evaluations. Many practical settings, however, allow only pairwise comparison queries, yielding a preferential BO problem where direct function evaluations are unavailable. Extending the knowledge gradient to preferential BO is hindered by its computational challenge. At its core, the look-ahead step in the preferential setting requires computing a non-Gaussian posterior, which was previously considered intractable. In this paper, we address this challenge by deriving an exact and analytical knowledge gradient for preferential BO. We show that the exact knowledge gradient performs strongly on a suite of benchmark problems, often outperforming existing acquisition functions. In addition, we also present a case study illustrating the limitation of the knowledge gradient in certain scenarios. △ Less

Submitted 29 January, 2026; originally announced January 2026.

VENI: Variational Encoder for Natural Illumination

Authors: Paul Walker, James A. D. Gardner, Andreea Ardelean, William A. P. Smith, Bernhard Egger

Abstract: Inverse rendering is an ill-posed problem, but priors such as illumination priors can help simplify it. Existing work either disregards the spherical and rotation-equivariant nature of illumination environments or does not provide a well-behaved latent space. We propose a rotation-equivariant variational autoencoder that models natural illumination on the sphere without relying on 2D projections.… ▽ More Inverse rendering is an ill-posed problem, but priors such as illumination priors can help simplify it. Existing work either disregards the spherical and rotation-equivariant nature of illumination environments or does not provide a well-behaved latent space. We propose a rotation-equivariant variational autoencoder that models natural illumination on the sphere without relying on 2D projections. To preserve the SO(2)-equivariance of environment maps, we use a novel Vector Neuron Vision Transformer (VN-ViT) as encoder and a rotation-equivariant conditional neural field as decoder. In the encoder, we reduce the equivariance from SO(3) to SO(2) using a novel SO(2)-equivariant fully connected layer, an extension of Vector Neurons. We show that our SO(2)-equivariant fully connected layer outperforms standard Vector Neurons when used in our SO(2)-equivariant model. Compared to previous methods, our variational autoencoder enables smoother interpolation in latent space and offers a more well-behaved latent space. △ Less

Submitted 24 June, 2026; v1 submitted 20 January, 2026; originally announced January 2026.

Comments: Project Repo - https://github.com/paul-pw/veni Project page - https://paul-pw.github.io/veni

Journal ref: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2026) 16248-16257

We Still Don't Understand High-Dimensional Bayesian Optimization

Authors: Colin Doumont, Donney Fan, Natalie Maus, Jacob R. Gardner, Henry Moss, Geoff Pleiss

Abstract: Existing high-dimensional Bayesian optimization (BO) methods aim to overcome the curse of dimensionality by carefully encoding structural assumptions, from locality to sparsity to smoothness, into the optimization procedure. Surprisingly, we demonstrate that these approaches are outperformed by arguably the simplest method imaginable: Bayesian linear regression. After applying a geometric transfor… ▽ More Existing high-dimensional Bayesian optimization (BO) methods aim to overcome the curse of dimensionality by carefully encoding structural assumptions, from locality to sparsity to smoothness, into the optimization procedure. Surprisingly, we demonstrate that these approaches are outperformed by arguably the simplest method imaginable: Bayesian linear regression. After applying a geometric transformation to avoid boundary-seeking behavior, Gaussian processes with linear kernels match state-of-the-art performance on tasks with 60- to 6,000-dimensional search spaces. Linear models offer numerous advantages over their non-parametric counterparts: they afford closed-form sampling and their computation scales linearly with data, a fact we exploit on molecular optimization tasks with >20,000 observations. Coupled with empirical analyses, our results suggest the need to depart from past intuitions about BO methods in high-dimensions. △ Less

Submitted 9 April, 2026; v1 submitted 28 November, 2025; originally announced December 2025.

TAPVid-360: Tracking Any Point in 360 from Narrow Field of View Video

Authors: Finlay G. C. Hudson, James A. D. Gardner, William A. P. Smith

Abstract: Humans excel at constructing panoramic mental models of their surroundings, maintaining object permanence and inferring scene structure beyond visible regions. In contrast, current artificial vision systems struggle with persistent, panoramic understanding, often processing scenes egocentrically on a frame-by-frame basis. This limitation is pronounced in the Track Any Point (TAP) task, where exist… ▽ More Humans excel at constructing panoramic mental models of their surroundings, maintaining object permanence and inferring scene structure beyond visible regions. In contrast, current artificial vision systems struggle with persistent, panoramic understanding, often processing scenes egocentrically on a frame-by-frame basis. This limitation is pronounced in the Track Any Point (TAP) task, where existing methods fail to track 2D points outside the field of view. To address this, we introduce TAPVid-360, a novel task that requires predicting the 3D direction to queried scene points across a video sequence, even when far outside the narrow field of view of the observed video. This task fosters learning allocentric scene representations without needing dynamic 4D ground truth scene models for training. Instead, we exploit 360 videos as a source of supervision, resampling them into narrow field-of-view perspectives while computing ground truth directions by tracking points across the full panorama using a 2D pipeline. We introduce a new dataset and benchmark, TAPVid360-10k comprising 10k perspective videos with ground truth directional point tracking. Our baseline adapts CoTracker v3 to predict per-point rotations for direction updates, outperforming existing TAP and TAPVid 3D methods. Project page: https://finlay-hudson.github.io/tapvid360 △ Less

Submitted 12 December, 2025; v1 submitted 26 November, 2025; originally announced November 2025.

Comments: Accepted to NeurIPS 2025

Data-Centric Lessons To Improve Speech-Language Pretraining

Authors: Vishaal Udandarao, Zhiyun Lu, Xuankai Chang, Yongqiang Wang, Violet Z. Yao, Albin Madapally Jose, Fartash Faghri, Josh Gardner, Chung-Cheng Chiu

Abstract: Spoken Question-Answering (SQA) is a core capability for useful and interactive artificial intelligence systems. Recently, several speech-language models (SpeechLMs) have been released with a specific focus on improving their SQA performance. However, a lack of controlled ablations of pretraining data processing and curation makes it challenging to understand what factors account for performance,… ▽ More Spoken Question-Answering (SQA) is a core capability for useful and interactive artificial intelligence systems. Recently, several speech-language models (SpeechLMs) have been released with a specific focus on improving their SQA performance. However, a lack of controlled ablations of pretraining data processing and curation makes it challenging to understand what factors account for performance, despite substantial gains from similar studies in other data modalities. In this work, we address this gap by conducting a data-centric exploration for pretraining SpeechLMs. We focus on three research questions fundamental to speech-language pretraining data: (1) how to process raw web-crawled audio content for speech-text pretraining, (2) how to construct synthetic pretraining datasets to augment web-crawled data and (3) how to interleave (text, audio) segments into training sequences. We apply the insights from our controlled data-centric ablations to pretrain a 3.8B-parameter SpeechLM, called SpeLangy, that outperforms models that are up to 3x larger by 10.2% absolute performance. We hope our findings highlight the impact of effective data curation for speech-language pretraining and guide future data-centric exploration in SpeechLMs. △ Less

Submitted 22 October, 2025; originally announced October 2025.

Comments: Tech Report

Bayesian Distributional Models of Executive Functioning

Authors: Robert Kasumba, Zeyu Lu, Dom CP Marticorena, Mingyang Zhong, Paul Beggs, Anja Pahor, Geetha Ramani, Imani Goffney, Susanne M Jaeggi, Aaron R Seitz, Jacob R Gardner, Dennis L Barbour

Abstract: This study uses controlled simulations with known ground-truth parameters to evaluate how Distributional Latent Variable Models (DLVM) and Bayesian Distributional Active LEarning (DALE) perform in comparison to conventional Independent Maximum Likelihood Estimation (IMLE). DLVM integrates observations across multiple executive function tasks and individuals, allowing parameter estimation even unde… ▽ More This study uses controlled simulations with known ground-truth parameters to evaluate how Distributional Latent Variable Models (DLVM) and Bayesian Distributional Active LEarning (DALE) perform in comparison to conventional Independent Maximum Likelihood Estimation (IMLE). DLVM integrates observations across multiple executive function tasks and individuals, allowing parameter estimation even under sparse or incomplete data conditions. To establish known-ground truth, we uniformly sample individual sessions from a neural network learned latent space and map them to distributional cognitive performance across different tasks. The individual test-items are then sampled from these distributions using either DALE, random procedure or a standard fixed battery approach. When given the same set of observations, DLVM consistently outperformed IMLE, especially under smaller amounts of data, and converges faster to highly accurate estimates of the true distributions. In a second set of analyses, DALE adaptively guided sampling to maximize information gain, outperforming random sampling and fixed test batteries, particularly within the first 80 trials. These findings establish the advantages of combining DLVM's cross-task inference with DALE's optimal adaptive sampling, providing a principled basis for more efficient cognitive assessments. △ Less

Submitted 13 August, 2026; v1 submitted 30 September, 2025; originally announced October 2025.

Comments: 45 pages, 9 figures, 2 tables + supplemental info

Is the `Agent' Paradigm a Limiting Framework for Next-Generation Intelligent Systems?

Authors: Jesse Gardner, Vladimir A. Baulin

Abstract: The concept of the 'agent' has profoundly shaped Artificial Intelligence (AI) research, guiding development from foundational theories to contemporary applications like Large Language Model (LLM)-based systems. This paper critically re-evaluates the necessity and optimality of this agent-centric paradigm. We argue that its persistent conceptual ambiguities and inherent anthropocentric biases may r… ▽ More The concept of the 'agent' has profoundly shaped Artificial Intelligence (AI) research, guiding development from foundational theories to contemporary applications like Large Language Model (LLM)-based systems. This paper critically re-evaluates the necessity and optimality of this agent-centric paradigm. We argue that its persistent conceptual ambiguities and inherent anthropocentric biases may represent a limiting framework. We distinguish between agentic systems (AI inspired by agency, often semi-autonomous, e.g., LLM-based agents), agential systems (fully autonomous, self-producing systems, currently only biological), and non-agentic systems (tools without the impression of agency). Our analysis, based on a systematic review of relevant literature, deconstructs the agent paradigm across various AI frameworks, highlighting challenges in defining and measuring properties like autonomy and goal-directedness. We argue that the 'agentic' framing of many AI systems, while heuristically useful, can be misleading and may obscure the underlying computational mechanisms, particularly in Large Language Models (LLMs). As an alternative, we propose a shift in focus towards frameworks grounded in system-level dynamics, world modeling, and material intelligence. We conclude that investigating non-agentic and systemic frameworks, inspired by complex systems, biology, and unconventional computing, is essential for advancing towards robust, scalable, and potentially non-anthropomorphic forms of general intelligence. This requires not only new architectures but also a fundamental reconsideration of our understanding of intelligence itself, moving beyond the agent metaphor. △ Less

Submitted 13 September, 2025; originally announced September 2025.

Greener Deep Reinforcement Learning: Analysis of Energy and Carbon Efficiency Across Atari Benchmarks

Authors: Jason Gardner, Ayan Dutta, Swapnoneel Roy, O. Patrick Kreidl, Ladislau Boloni

Abstract: The growing computational demands of deep reinforcement learning (DRL) have raised concerns about the environmental and economic costs of training large-scale models. While algorithmic efficiency in terms of learning performance has been extensively studied, the energy requirements, greenhouse gas emissions, and monetary costs of DRL algorithms remain largely unexplored. In this work, we present a… ▽ More The growing computational demands of deep reinforcement learning (DRL) have raised concerns about the environmental and economic costs of training large-scale models. While algorithmic efficiency in terms of learning performance has been extensively studied, the energy requirements, greenhouse gas emissions, and monetary costs of DRL algorithms remain largely unexplored. In this work, we present a systematic benchmarking study of the energy consumption of seven state-of-the-art DRL algorithms, namely DQN, TRPO, A2C, ARS, PPO, RecurrentPPO, and QR-DQN, implemented using Stable Baselines. Each algorithm was trained for one million steps each on ten Atari 2600 games, and power consumption was measured in real-time to estimate total energy usage, CO2-Equivalent emissions, and electricity cost based on the U.S. national average electricity price. Our results reveal substantial variation in energy efficiency and training cost across algorithms, with some achieving comparable performance while consuming up to 24% less energy (ARS vs. DQN), emitting nearly 68% less CO2, and incurring almost 68% lower monetary cost (QR-DQN vs. RecurrentPPO) than less efficient counterparts. We further analyze the trade-offs between learning performance, training time, energy use, and financial cost, highlighting cases where algorithmic choices can mitigate environmental and economic impact without sacrificing learning performance. This study provides actionable insights for developing energy-aware and cost-efficient DRL practices and establishes a foundation for incorporating sustainability considerations into future algorithmic design and evaluation. △ Less

Submitted 5 September, 2025; originally announced September 2025.

Comments: Submitted to a journal - under review

OLMoASR: Open Models and Data for Training Robust Speech Recognition Models

Authors: Huong Ngo, Matt Deitke, Martijn Bartelds, Sarah Pratt, Josh Gardner, Matt Jordan, Ludwig Schmidt

Abstract: Improvements in training data scale and quality have led to significant advances, yet its influence in speech recognition remains underexplored. In this paper, we present a large-scale dataset, OLMoASR-Pool, and series of models, OLMoASR, to study and develop robust zero-shot speech recognition models. Beginning from OLMoASR-Pool, a collection of 3M hours of English audio and 17M transcripts, we d… ▽ More Improvements in training data scale and quality have led to significant advances, yet its influence in speech recognition remains underexplored. In this paper, we present a large-scale dataset, OLMoASR-Pool, and series of models, OLMoASR, to study and develop robust zero-shot speech recognition models. Beginning from OLMoASR-Pool, a collection of 3M hours of English audio and 17M transcripts, we design text heuristic filters to remove low-quality or mistranscribed data. Our curation pipeline produces a new dataset containing 1M hours of high-quality audio-transcript pairs, which we call OLMoASR-Mix. We use OLMoASR-Mix to train the OLMoASR-Mix suite of models, ranging from 39M (tiny.en) to 1.5B (large.en) parameters. Across all model scales, OLMoASR achieves comparable average performance to OpenAI's Whisper on short and long-form speech recognition benchmarks. Notably, OLMoASR-medium.en attains a 12.8\% and 11.0\% word error rate (WER) that is on par with Whisper's largest English-only model Whisper-medium.en's 12.4\% and 10.5\% WER for short and long-form recognition respectively (at equivalent parameter count). OLMoASR-Pool, OLMoASR models, and filtering, training and evaluation code will be made publicly available to further research on robust speech processing. △ Less

Submitted 28 August, 2025; originally announced August 2025.

Comments: 17 pages, 7 figures

A Dataset for Distilling Knowledge Priors from Literature for Therapeutic Design

Authors: Haydn Thomas Jones, Natalie Maus, Josh Magnus Ludan, Maggie Ziyu Huan, Jiaming Liang, Marcelo Der Torossian Torres, Jiatao Liang, Zachary Ives, Yoseph Barash, Cesar de la Fuente-Nunez, Jacob R. Gardner, Mark Yatskar

Abstract: AI-driven discovery can greatly reduce design time and enhance new therapeutics' effectiveness. Models using simulators explore broad design spaces but risk violating implicit constraints due to a lack of experimental priors. For example, in a new analysis we performed on a diverse set of models on the GuacaMol benchmark using supervised classifiers, over 60\% of molecules proposed had high probab… ▽ More AI-driven discovery can greatly reduce design time and enhance new therapeutics' effectiveness. Models using simulators explore broad design spaces but risk violating implicit constraints due to a lack of experimental priors. For example, in a new analysis we performed on a diverse set of models on the GuacaMol benchmark using supervised classifiers, over 60\% of molecules proposed had high probability of being mutagenic. In this work, we introduce Medex, a dataset of priors for design problems extracted from literature describing compounds used in lab settings. It is constructed with LLM pipelines for discovering therapeutic entities in relevant paragraphs and summarizing information in concise fair-use facts. Medex consists of 32.3 million pairs of natural language facts, and appropriate entity representations (i.e. SMILES or refseq IDs). To demonstrate the potential of the data, we train LLM, CLIP, and LLava architectures to reason jointly about text and design targets and evaluate on tasks from the Therapeutic Data Commons (TDC). Medex is highly effective for creating models with strong priors: in supervised prediction problems that use our data as pretraining, our best models with 15M learnable parameters outperform larger 2B TxGemma on both regression and classification TDC tasks, and perform comparably to 9B models on average. Models built with Medex can be used as constraints while optimizing for novel molecules in GuacaMol, resulting in proposals that are safer and nearly as effective. We release our dataset at https://huggingface.co/datasets/medexanon/Medex, and will provide expanded versions as available literature grows. △ Less

Submitted 11 September, 2025; v1 submitted 14 August, 2025; originally announced August 2025.

Apple Intelligence Foundation Language Models: Tech Report 2025

Authors: Ethan Li, Anders Boesen Lindbo Larsen, Chen Zhang, Xiyou Zhou, Jun Qin, Dian Ang Yap, Narendran Raghavan, Xuankai Chang, Margit Bowler, Eray Yildiz, John Peebles, Hannah Gillis Coleman, Matteo Ronchi, Peter Gray, Keen You, Anthony Spalvieri-Kruse, Ruoming Pang, Reed Li, Yuli Yang, Emad Soroush, Zhiyun Lu, Crystal Xiao, Rong Situ, Jordan Huffaker, David Griffiths , et al. (373 additional authors not shown)

Abstract: We introduce two multilingual, multimodal foundation language models that power Apple Intelligence features across Apple devices and services: i a 3B-parameter on-device model optimized for Apple silicon through architectural innovations such as KV-cache sharing and 2-bit quantization-aware training; and ii a scalable server model built on a novel Parallel-Track Mixture-of-Experts PT-MoE transform… ▽ More We introduce two multilingual, multimodal foundation language models that power Apple Intelligence features across Apple devices and services: i a 3B-parameter on-device model optimized for Apple silicon through architectural innovations such as KV-cache sharing and 2-bit quantization-aware training; and ii a scalable server model built on a novel Parallel-Track Mixture-of-Experts PT-MoE transformer that combines track parallelism, mixture-of-experts sparse computation, and interleaved global-local attention to deliver high quality with competitive cost on Apple's Private Cloud Compute platform. Both models are trained on large-scale multilingual and multimodal datasets sourced via responsible web crawling, licensed corpora, and high-quality synthetic data, then further refined with supervised fine-tuning and reinforcement learning on a new asynchronous platform. The resulting models support several additional languages while understanding images and executing tool calls. In public benchmarks and human evaluations, both the server model and the on-device model match or surpass comparably sized open baselines. A new Swift-centric Foundation Models framework exposes guided generation, constrained tool calling, and LoRA adapter fine-tuning, allowing developers to integrate these capabilities with a few lines of code. The latest advancements in Apple Intelligence models are grounded in our Responsible AI approach with safeguards like content filtering and locale-specific evaluation, as well as our commitment to protecting our users' privacy with innovations like Private Cloud Compute. △ Less

Submitted 27 August, 2025; v1 submitted 17 July, 2025; originally announced July 2025.

Language Models Improve When Pretraining Data Matches Target Tasks

Authors: David Mizrahi, Anders Boesen Lindbo Larsen, Jesse Allardice, Suzie Petryk, Yuri Gorokhov, Jeffrey Li, Alex Fang, Josh Gardner, Tom Gunter, Afshin Dehghan

Abstract: Every data selection method inherently has a target. In practice, these targets often emerge implicitly through benchmark-driven iteration: researchers develop selection strategies, train models, measure benchmark performance, then refine accordingly. This raises a natural question: what happens when we make this optimization explicit? To explore this, we propose benchmark-targeted ranking (BETR),… ▽ More Every data selection method inherently has a target. In practice, these targets often emerge implicitly through benchmark-driven iteration: researchers develop selection strategies, train models, measure benchmark performance, then refine accordingly. This raises a natural question: what happens when we make this optimization explicit? To explore this, we propose benchmark-targeted ranking (BETR), a simple method that selects pretraining documents based on similarity to benchmark training examples. BETR embeds benchmark examples and a sample of pretraining documents in a shared space, scores this sample by similarity to benchmarks, then trains a lightweight classifier to predict these scores for the full corpus. We compare data selection methods by training over 500 models spanning $10^{19}$ to $10^{22}$ FLOPs and fitting scaling laws to them. From this, we find that simply aligning pretraining data to evaluation benchmarks using BETR achieves a 2.1x compute multiplier over DCLM-Baseline (4.7x over unfiltered data) and improves performance on 9 out of 10 tasks across all scales. BETR also generalizes well: when targeting a diverse set of benchmarks disjoint from our evaluation suite, it still matches or outperforms baselines. Our scaling analysis further reveals a clear trend: larger models require less aggressive filtering. Overall, our findings show that directly matching pretraining data to target tasks precisely shapes model capabilities and highlight that optimal selection strategies must adapt to model scale. △ Less

Submitted 16 July, 2025; originally announced July 2025.

Comments: 44 pages, 25 figures, 13 tables

New Time Integrators and Capabilities in SUNDIALS Versions 6.2.0-7.4.0

Authors: Steven B. Roberts, Mustafa Ağgül, Daniel R. Reynolds, Cody J. Balos, David J. Gardner, Carol S. Woodward

Abstract: SUNDIALS is a well-established numerical library that provides robust and efficient time integrators and nonlinear solvers. This paper overviews several significant improvements and new features added over the last three years to support scientific simulations run on high-performance computing systems. Notably, three new classes of one-step methods have been implemented: low storage Runge-Kutta, s… ▽ More SUNDIALS is a well-established numerical library that provides robust and efficient time integrators and nonlinear solvers. This paper overviews several significant improvements and new features added over the last three years to support scientific simulations run on high-performance computing systems. Notably, three new classes of one-step methods have been implemented: low storage Runge-Kutta, symplectic partitioned Runge-Kutta, and operator splitting. In addition, we describe new time step adaptivity support for multirate methods, adjoint sensitivity analysis capabilities for explicit Runge-Kutta methods, additional options for Anderson acceleration in nonlinear solvers, and improved error handling and logging. △ Less

Submitted 9 February, 2026; v1 submitted 17 June, 2025; originally announced June 2025.

UTSA-NLP at ArchEHR-QA 2025: Improving EHR Question Answering via Self-Consistency Prompting

Authors: Sara Shields-Menard, Zach Reimers, Joshua Gardner, David Perry, Anthony Rios

Abstract: We describe our system for the ArchEHR-QA Shared Task on answering clinical questions using electronic health records (EHRs). Our approach uses large language models in two steps: first, to find sentences in the EHR relevant to a clinician's question, and second, to generate a short, citation-supported response based on those sentences. We use few-shot prompting, self-consistency, and thresholding… ▽ More We describe our system for the ArchEHR-QA Shared Task on answering clinical questions using electronic health records (EHRs). Our approach uses large language models in two steps: first, to find sentences in the EHR relevant to a clinician's question, and second, to generate a short, citation-supported response based on those sentences. We use few-shot prompting, self-consistency, and thresholding to improve the sentence classification step to decide which sentences are essential. We compare several models and find that a smaller 8B model performs better than a larger 70B model for identifying relevant information. Our results show that accurate sentence selection is critical for generating high-quality responses and that self-consistency with thresholding helps make these decisions more reliable. △ Less

Submitted 5 June, 2025; originally announced June 2025.

Comments: Accepted to BioNLP 2025

Nearly Dimension-Independent Convergence of Mean-Field Black-Box Variational Inference

Authors: Kyurae Kim, Yi-An Ma, Trevor Campbell, Jacob R. Gardner

Abstract: We prove that, given a mean-field location-scale variational family, black-box variational inference (BBVI) with the reparametrization gradient converges at a rate that is nearly independent of explicit dimension dependence. Specifically, for a $d$-dimensional strongly log-concave and log-smooth target, the number of iterations for BBVI with a sub-Gaussian family to obtain a solution $ε$-close to… ▽ More We prove that, given a mean-field location-scale variational family, black-box variational inference (BBVI) with the reparametrization gradient converges at a rate that is nearly independent of explicit dimension dependence. Specifically, for a $d$-dimensional strongly log-concave and log-smooth target, the number of iterations for BBVI with a sub-Gaussian family to obtain a solution $ε$-close to the global optimum has a dimension dependence of $\mathrm{O}(\log d)$. This is a significant improvement over the $\mathrm{O}(d)$ dependence of full-rank location-scale families. For heavy-tailed families, we prove a weaker $\mathrm{O}(d^{2/k})$ dependence, where $k$ is the number of finite moments of the family. Additionally, if the Hessian of the target log-density is constant, the complexity is free of any explicit dimension dependence. We also prove that our bound on the gradient variance, which is key to our result, cannot be improved using only spectral bounds on the Hessian of the target log-density. △ Less

Submitted 20 October, 2025; v1 submitted 27 May, 2025; originally announced May 2025.

Comments: Accepted to NeurIPS'25

NeuRaLaTeX: A machine learning library written in pure LaTeX

Authors: James A. D. Gardner, Will Rowan, William A. P. Smith

Abstract: In this paper, we introduce NeuRaLaTeX, which we believe to be the first deep learning library written entirely in LaTeX. As part of your LaTeX document you can specify the architecture of a neural network and its loss functions, define how to generate or load training data, and specify training hyperparameters and experiments. When the document is compiled, the LaTeX compiler will generate or loa… ▽ More In this paper, we introduce NeuRaLaTeX, which we believe to be the first deep learning library written entirely in LaTeX. As part of your LaTeX document you can specify the architecture of a neural network and its loss functions, define how to generate or load training data, and specify training hyperparameters and experiments. When the document is compiled, the LaTeX compiler will generate or load training data, train the network, run experiments, and generate figures. This paper generates a random 100 point spiral dataset, trains a two layer MLP on it, evaluates on a different random spiral dataset, produces plots and tables of results. The paper took 48 hours to compile and the entire source code for NeuRaLaTeX is contained within the source code of the paper. We propose two new metrics: the Written In Latex (WIL) metric measures the proportion of a machine learning library that is written in pure LaTeX, while the Source Code Of Method in Source Code of Paper (SCOMISCOP) metric measures the proportion of a paper's implementation that is contained within the paper source. We are state-of-the-art for both metrics, outperforming the ResNet and Transformer papers, as well as the PyTorch and Tensorflow libraries. Source code, documentation, videos, crypto scams and an invitation to invest in the commercialisation of NeuRaLaTeX are available at https://www.neuralatex.com △ Less

Submitted 2 April, 2025; v1 submitted 31 March, 2025; originally announced March 2025.

Tuning Sequential Monte Carlo Samplers via Greedy Incremental Divergence Minimization

Authors: Kyurae Kim, Zuheng Xu, Jacob R. Gardner, Trevor Campbell

Abstract: The performance of sequential Monte Carlo (SMC) samplers heavily depends on the tuning of the Markov kernels used in the path proposal. For SMC samplers with unadjusted Markov kernels, standard tuning objectives, such as the Metropolis-Hastings acceptance rate or the expected-squared jump distance, are no longer applicable. While stochastic gradient-based end-to-end optimization has been explored… ▽ More The performance of sequential Monte Carlo (SMC) samplers heavily depends on the tuning of the Markov kernels used in the path proposal. For SMC samplers with unadjusted Markov kernels, standard tuning objectives, such as the Metropolis-Hastings acceptance rate or the expected-squared jump distance, are no longer applicable. While stochastic gradient-based end-to-end optimization has been explored for tuning SMC samplers, they often incur excessive training costs, even for tuning just the kernel step sizes. In this work, we propose a general adaptation framework for tuning the Markov kernels in SMC samplers by minimizing the incremental Kullback-Leibler (KL) divergence between the proposal and target paths. For step size tuning, we provide a gradient- and tuning-free algorithm that is generally applicable for kernels such as Langevin Monte Carlo (LMC). We further demonstrate the utility of our approach by providing a tailored scheme for tuning kinetic LMC used in SMC samplers. Our implementations are able to obtain a full schedule of tuned parameters at the cost of a few vanilla SMC runs, which is a fraction of gradient-based approaches. △ Less

Submitted 9 December, 2025; v1 submitted 19 March, 2025; originally announced March 2025.

Comments: Accepted to ICML'25; v4, v5: fixed typos; v6: fixed missing log in display after Eq (8)

Large Scale Multi-Task Bayesian Optimization with Large Language Models

Authors: Yimeng Zeng, Natalie Maus, Haydn Thomas Jones, Jeffrey Tao, Fangping Wan, Marcelo Der Torossian Torres, Cesar de la Fuente-Nunez, Ryan Marcus, Osbert Bastani, Jacob R. Gardner

Abstract: In multi-task Bayesian optimization, the goal is to leverage experience from optimizing existing tasks to improve the efficiency of optimizing new ones. While approaches using multi-task Gaussian processes or deep kernel transfer exist, the performance improvement is marginal when scaling beyond a moderate number of tasks. We introduce a novel approach leveraging large language models (LLMs) to le… ▽ More In multi-task Bayesian optimization, the goal is to leverage experience from optimizing existing tasks to improve the efficiency of optimizing new ones. While approaches using multi-task Gaussian processes or deep kernel transfer exist, the performance improvement is marginal when scaling beyond a moderate number of tasks. We introduce a novel approach leveraging large language models (LLMs) to learn from, and improve upon, previous optimization trajectories, scaling to approximately 1500 distinct tasks. Specifically, we propose a feedback loop in which an LLM is fine-tuned on the high quality solutions to specific tasks found by Bayesian optimization (BO). This LLM is then used to generate initialization points for future BO searches for new tasks. The trajectories of these new searches provide additional training data for fine-tuning the LLM, completing the loop. We evaluate our method on two distinct domains: database query optimization and antimicrobial peptide design. Results demonstrate that our approach creates a positive feedback loop, where the LLM's generated initializations gradually improve, leading to better optimization performance. As this feedback loop continues, we find that the LLM is eventually able to generate solutions to new tasks in just a few shots that are better than the solutions produced by "from scratch" by Bayesian optimization while simultaneously requiring significantly fewer oracle calls. △ Less

Submitted 12 June, 2025; v1 submitted 11 March, 2025; originally announced March 2025.

Learned Offline Query Planning via Bayesian Optimization

Authors: Jeffrey Tao, Natalie Maus, Haydn Jones, Yimeng Zeng, Jacob R. Gardner, Ryan Marcus

Abstract: Analytics database workloads often contain queries that are executed repeatedly. Existing optimization techniques generally prioritize keeping optimization cost low, normally well below the time it takes to execute a single instance of a query. If a given query is going to be executed thousands of times, could it be worth investing significantly more optimization time? In contrast to traditional o… ▽ More Analytics database workloads often contain queries that are executed repeatedly. Existing optimization techniques generally prioritize keeping optimization cost low, normally well below the time it takes to execute a single instance of a query. If a given query is going to be executed thousands of times, could it be worth investing significantly more optimization time? In contrast to traditional online query optimizers, we propose an offline query optimizer that searches a wide variety of plans and incorporates query execution as a primitive. Our offline query optimizer combines variational auto-encoders with Bayesian optimization to find optimized plans for a given query. We compare our technique to the optimal plans possible with PostgreSQL and recent RL-based systems over several datasets, and show that our technique finds faster query plans. △ Less

Submitted 7 February, 2025; originally announced February 2025.

Covering Multiple Objectives with a Small Set of Solutions Using Bayesian Optimization

Authors: Natalie Maus, Kyurae Kim, Yimeng Zeng, Haydn Thomas Jones, Fangping Wan, Marcelo Der Torossian Torres, Cesar de la Fuente-Nunez, Jacob R. Gardner

Abstract: In multi-objective black-box optimization, the goal is typically to find solutions that optimize a set of $T$ black-box objective functions, $f_1, \ldots f_T$, simultaneously. Traditional approaches often seek a single Pareto-optimal set that balances trade-offs among all objectives. In contrast, we consider a problem setting that departs from this paradigm: finding a small set of $K < T$ solution… ▽ More In multi-objective black-box optimization, the goal is typically to find solutions that optimize a set of $T$ black-box objective functions, $f_1, \ldots f_T$, simultaneously. Traditional approaches often seek a single Pareto-optimal set that balances trade-offs among all objectives. In contrast, we consider a problem setting that departs from this paradigm: finding a small set of $K < T$ solutions, that collectively "cover" the $T$ objectives. A set of solutions is defined as "covering" if, for each objective $f_1, \ldots f_T$, there is at least one good solution. A motivating example for this problem setting occurs in drug design. For example, we may have $T$ pathogens and aim to identify a set of $K < T$ antibiotics such that at least one antibiotic can be used to treat each pathogen. This problem, known as coverage optimization, has yet to be tackled with the Bayesian optimization (BO) framework. To fill this void, we develop Multi-Objective Coverage Bayesian Optimization (MOCOBO), a BO algorithm for solving coverage optimization. Our approach is based on a new acquisition function reminiscent of expected improvement in the vanilla BO setup. We demonstrate the performance of our method on high-dimensional black-box optimization tasks, including applications in peptide and molecular design. Results show that the coverage of the $K < T$ solutions found by MOCOBO matches or nearly matches the coverage of $T$ solutions obtained by optimizing each objective individually. Furthermore, in in vitro experiments, the peptides found by MOCOBO exhibited high potency against drug-resistant pathogens, further demonstrating the potential of MOCOBO for drug discovery. All of our code is publicly available at the following link: https://github.com/nataliemaus/mocobo. △ Less

Submitted 27 October, 2025; v1 submitted 31 January, 2025; originally announced January 2025.

Improving Satellite Imagery Masking using Multi-task and Transfer Learning

Authors: Rangel Daroya, Luisa Vieira Lucchese, Travis Simmons, Punwath Prum, Tamlin Pavelsky, John Gardner, Colin J. Gleason, Subhransu Maji

Abstract: Many remote sensing applications employ masking of pixels in satellite imagery for subsequent measurements. For example, estimating water quality variables, such as Suspended Sediment Concentration (SSC) requires isolating pixels depicting water bodies unaffected by clouds, their shadows, terrain shadows, and snow and ice formation. A significant bottleneck is the reliance on a variety of data pro… ▽ More Many remote sensing applications employ masking of pixels in satellite imagery for subsequent measurements. For example, estimating water quality variables, such as Suspended Sediment Concentration (SSC) requires isolating pixels depicting water bodies unaffected by clouds, their shadows, terrain shadows, and snow and ice formation. A significant bottleneck is the reliance on a variety of data products (e.g., satellite imagery, elevation maps), and a lack of precision in individual steps affecting estimation accuracy. We propose to improve both the accuracy and computational efficiency of masking by developing a system that predicts all required masks from Harmonized Landsat and Sentinel (HLS) imagery. Our model employs multi-tasking to share computation and enable higher accuracy across tasks. We experiment with recent advances in deep network architectures and show that masking models can benefit from these, especially when combined with pre-training on large satellite imagery datasets. We present a collection of models offering different speed/accuracy trade-offs for masking. MobileNet variants are the fastest, and perform competitively with larger architectures. Transformer-based architectures are the slowest, but benefit the most from pre-training on large satellite imagery datasets. Our models provide a 9% F1 score improvement compared to previous work on water pixel identification. When integrated with an SSC estimation system, our models result in a 30x speedup while reducing estimation error by 2.64 mg/L, allowing for global-scale analysis. We also evaluate our model on a recently proposed cloud and cloud shadow estimation benchmark, where we outperform the current state-of-the-art model by at least 6% in F1 score. △ Less

Submitted 11 December, 2024; originally announced December 2024.

Computation-Aware Gaussian Processes: Model Selection And Linear-Time Inference

Authors: Jonathan Wenger, Kaiwen Wu, Philipp Hennig, Jacob R. Gardner, Geoff Pleiss, John P. Cunningham

Abstract: Model selection in Gaussian processes scales prohibitively with the size of the training dataset, both in time and memory. While many approximations exist, all incur inevitable approximation error. Recent work accounts for this error in the form of computational uncertainty, which enables -- at the cost of quadratic complexity -- an explicit tradeoff between computation and precision. Here we exte… ▽ More Model selection in Gaussian processes scales prohibitively with the size of the training dataset, both in time and memory. While many approximations exist, all incur inevitable approximation error. Recent work accounts for this error in the form of computational uncertainty, which enables -- at the cost of quadratic complexity -- an explicit tradeoff between computation and precision. Here we extend this development to model selection, which requires significant enhancements to the existing approach, including linear-time scaling in the size of the dataset. We propose a novel training loss for hyperparameter optimization and demonstrate empirically that the resulting method can outperform SGPR, CGGP and SVGP, state-of-the-art methods for GP model selection, on medium to large-scale datasets. Our experiments show that model selection for computation-aware GPs trained on 1.8 million data points can be done within a few hours on a single GPU. As a result of this work, Gaussian processes can be trained on large-scale datasets without significantly compromising their ability to quantify uncertainty -- a fundamental prerequisite for optimal decision-making. △ Less

Submitted 7 July, 2025; v1 submitted 1 November, 2024; originally announced November 2024.

Comments: Advances in Neural Information Processing Systems (NeurIPS 2024)

Improving Structural Diversity of Blackbox LLMs via Chain-of-Specification Prompting

Authors: Halley Young, Yimeng Zeng, Jacob Gardner, Osbert Bastani

Abstract: The capability to generate diverse text is a key challenge facing large language models (LLMs). Thus far, diversity has been studied via metrics such as $n$-gram diversity or diversity of BERT embeddings. However, for these kinds of diversity, the user has little control over the dimensions along which diversity is considered. For example, in the poetry domain, one might desire diversity in terms… ▽ More The capability to generate diverse text is a key challenge facing large language models (LLMs). Thus far, diversity has been studied via metrics such as $n$-gram diversity or diversity of BERT embeddings. However, for these kinds of diversity, the user has little control over the dimensions along which diversity is considered. For example, in the poetry domain, one might desire diversity in terms of rhyme and meter, whereas in the code domain, one might desire diversity in terms of the kinds of expressions used to solve a problem. We propose a diversity metric called structural diversity, where the user provides a mapping from generated text to features capturing the kinds of diversity that they care about. In addition, we propose a novel strategy called chain-of-specification (CoS) prompting for improving diversity by first having the LLM generate a specification encoding one instance of structural features, and then prompting the LLM to generate text that satisfies these features; notably, our strategy works with blackbox LLMs. In our experiments, we show that for structural diversity in the poetry and code domains, CoS significantly improves diversity compared to several baselines. △ Less

Submitted 12 August, 2024; originally announced August 2024.

A Fast, Robust Elliptical Slice Sampling Implementation for Linearly Truncated Multivariate Normal Distributions

Authors: Kaiwen Wu, Jacob R. Gardner

Abstract: Elliptical slice sampling, when adapted to linearly truncated multivariate normal distributions, is a rejection-free Markov chain Monte Carlo method. At its core, it requires analytically constructing an ellipse-polytope intersection. The main novelty of this paper is an algorithm that computes this intersection in $\mathcal{O}(m \log m)$ time, where $m$ is the number of linear inequality constrai… ▽ More Elliptical slice sampling, when adapted to linearly truncated multivariate normal distributions, is a rejection-free Markov chain Monte Carlo method. At its core, it requires analytically constructing an ellipse-polytope intersection. The main novelty of this paper is an algorithm that computes this intersection in $\mathcal{O}(m \log m)$ time, where $m$ is the number of linear inequality constraints representing the polytope. We show that an implementation based on this algorithm enhances numerical stability, speeds up running time, and is easy to parallelize for launching multiple Markov chains. △ Less

Submitted 15 July, 2024; originally announced July 2024.

Comments: 13 pages

Large Scale Transfer Learning for Tabular Data via Language Modeling

Authors: Josh Gardner, Juan C. Perdomo, Ludwig Schmidt

Abstract: Tabular data -- structured, heterogeneous, spreadsheet-style data with rows and columns -- is widely used in practice across many domains. However, while recent foundation models have reduced the need for developing task-specific datasets and predictors in domains such as language modeling and computer vision, this transfer learning paradigm has not had similar impact in the tabular domain. In thi… ▽ More Tabular data -- structured, heterogeneous, spreadsheet-style data with rows and columns -- is widely used in practice across many domains. However, while recent foundation models have reduced the need for developing task-specific datasets and predictors in domains such as language modeling and computer vision, this transfer learning paradigm has not had similar impact in the tabular domain. In this work, we seek to narrow this gap and present TabuLa-8B, a language model for tabular prediction. We define a process for extracting a large, high-quality training dataset from the TabLib corpus, proposing methods for tabular data filtering and quality control. Using the resulting dataset, which comprises over 2.1B rows from over 4M unique tables, we fine-tune a Llama 3-8B large language model (LLM) for tabular data prediction (classification and binned regression) using a novel packing and attention scheme for tabular prediction. Through evaluation across a test suite of 329 datasets, we find that TabuLa-8B has zero-shot accuracy on unseen tables that is over 15 percentage points (pp) higher than random guessing, a feat that is not possible with existing state-of-the-art tabular prediction models (e.g. XGBoost, TabPFN). In the few-shot setting (1-32 shots), without any fine-tuning on the target datasets, TabuLa-8B is 5-15 pp more accurate than XGBoost and TabPFN models that are explicitly trained on equal, or even up to 16x more data. We release our model, code, and data along with the publication of this paper. △ Less

Submitted 20 November, 2024; v1 submitted 17 June, 2024; originally announced June 2024.

Comments: NeurIPS 2024 camera-ready updates

DataComp-LM: In search of the next generation of training sets for language models

Authors: Jeffrey Li, Alex Fang, Georgios Smyrnis, Maor Ivgi, Matt Jordan, Samir Gadre, Hritik Bansal, Etash Guha, Sedrick Keh, Kushal Arora, Saurabh Garg, Rui Xin, Niklas Muennighoff, Reinhard Heckel, Jean Mercat, Mayee Chen, Suchin Gururangan, Mitchell Wortsman, Alon Albalak, Yonatan Bitton, Marianna Nezhurina, Amro Abbas, Cheng-Yu Hsieh, Dhruba Ghosh, Josh Gardner , et al. (34 additional authors not shown)

Abstract: We introduce DataComp for Language Models (DCLM), a testbed for controlled dataset experiments with the goal of improving language models. As part of DCLM, we provide a standardized corpus of 240T tokens extracted from Common Crawl, effective pretraining recipes based on the OpenLM framework, and a broad suite of 53 downstream evaluations. Participants in the DCLM benchmark can experiment with dat… ▽ More We introduce DataComp for Language Models (DCLM), a testbed for controlled dataset experiments with the goal of improving language models. As part of DCLM, we provide a standardized corpus of 240T tokens extracted from Common Crawl, effective pretraining recipes based on the OpenLM framework, and a broad suite of 53 downstream evaluations. Participants in the DCLM benchmark can experiment with data curation strategies such as deduplication, filtering, and data mixing at model scales ranging from 412M to 7B parameters. As a baseline for DCLM, we conduct extensive experiments and find that model-based filtering is key to assembling a high-quality training set. The resulting dataset, DCLM-Baseline enables training a 7B parameter language model from scratch to 64% 5-shot accuracy on MMLU with 2.6T training tokens. Compared to MAP-Neo, the previous state-of-the-art in open-data language models, DCLM-Baseline represents a 6.6 percentage point improvement on MMLU while being trained with 40% less compute. Our baseline model is also comparable to Mistral-7B-v0.3 and Llama 3 8B on MMLU (63% & 66%), and performs similarly on an average of 53 natural language understanding tasks while being trained with 6.6x less compute than Llama 3 8B. Our results highlight the importance of dataset design for training language models and offer a starting point for further research on data curation. △ Less

Submitted 21 April, 2025; v1 submitted 17 June, 2024; originally announced June 2024.

Comments: Project page: https://www.datacomp.ai/dclm/

Approximation-Aware Bayesian Optimization

Authors: Natalie Maus, Kyurae Kim, Geoff Pleiss, David Eriksson, John P. Cunningham, Jacob R. Gardner

Abstract: High-dimensional Bayesian optimization (BO) tasks such as molecular design often require 10,000 function evaluations before obtaining meaningful results. While methods like sparse variational Gaussian processes (SVGPs) reduce computational requirements in these settings, the underlying approximations result in suboptimal data acquisitions that slow the progress of optimization. In this paper we mo… ▽ More High-dimensional Bayesian optimization (BO) tasks such as molecular design often require 10,000 function evaluations before obtaining meaningful results. While methods like sparse variational Gaussian processes (SVGPs) reduce computational requirements in these settings, the underlying approximations result in suboptimal data acquisitions that slow the progress of optimization. In this paper we modify SVGPs to better align with the goals of BO: targeting informed data acquisition rather than global posterior fidelity. Using the framework of utility-calibrated variational inference, we unify GP approximation and data acquisition into a joint optimization problem, thereby ensuring optimal decisions under a limited computational budget. Our approach can be used with any decision-theoretic acquisition function and is compatible with trust region methods like TuRBO. We derive efficient joint objectives for the expected improvement and knowledge gradient acquisition functions in both the standard and batch BO settings. Our approach outperforms standard SVGPs on high-dimensional benchmark tasks in control and molecular design. △ Less

Submitted 9 June, 2025; v1 submitted 6 June, 2024; originally announced June 2024.

Zeroth-Order Fine-Tuning of LLMs with Extreme Sparsity

Authors: Wentao Guo, Jikai Long, Yimeng Zeng, Zirui Liu, Xinyu Yang, Yide Ran, Jacob R. Gardner, Osbert Bastani, Christopher De Sa, Xiaodong Yu, Beidi Chen, Zhaozhuo Xu

Abstract: Zeroth-order optimization (ZO) is a memory-efficient strategy for fine-tuning Large Language Models using only forward passes. However, the application of ZO fine-tuning in memory-constrained settings such as mobile phones and laptops is still challenging since full precision forward passes are infeasible. In this study, we address this limitation by integrating sparsity and quantization into ZO f… ▽ More Zeroth-order optimization (ZO) is a memory-efficient strategy for fine-tuning Large Language Models using only forward passes. However, the application of ZO fine-tuning in memory-constrained settings such as mobile phones and laptops is still challenging since full precision forward passes are infeasible. In this study, we address this limitation by integrating sparsity and quantization into ZO fine-tuning of LLMs. Specifically, we investigate the feasibility of fine-tuning an extremely small subset of LLM parameters using ZO. This approach allows the majority of un-tuned parameters to be quantized to accommodate the constraint of limited device memory. Our findings reveal that the pre-training process can identify a set of "sensitive parameters" that can guide the ZO fine-tuning of LLMs on downstream tasks. Our results demonstrate that fine-tuning 0.1% sensitive parameters in the LLM with ZO can outperform the full ZO fine-tuning performance, while offering wall-clock time speedup. Additionally, we show that ZO fine-tuning targeting these 0.1% sensitive parameters, combined with 4 bit quantization, enables efficient ZO fine-tuning of an Llama2-7B model on a GPU device with less than 8 GiB of memory and notably reduced latency. △ Less

Submitted 5 June, 2024; originally announced June 2024.

Understanding Stochastic Natural Gradient Variational Inference

Authors: Kaiwen Wu, Jacob R. Gardner

Abstract: Stochastic natural gradient variational inference (NGVI) is a popular posterior inference method with applications in various probabilistic models. Despite its wide usage, little is known about the non-asymptotic convergence rate in the \emph{stochastic} setting. We aim to lessen this gap and provide a better understanding. For conjugate likelihoods, we prove the first $\mathcal{O}(\frac{1}{T})$ n… ▽ More Stochastic natural gradient variational inference (NGVI) is a popular posterior inference method with applications in various probabilistic models. Despite its wide usage, little is known about the non-asymptotic convergence rate in the \emph{stochastic} setting. We aim to lessen this gap and provide a better understanding. For conjugate likelihoods, we prove the first $\mathcal{O}(\frac{1}{T})$ non-asymptotic convergence rate of stochastic NGVI. The complexity is no worse than stochastic gradient descent (\aka black-box variational inference) and the rate likely has better constant dependency that leads to faster convergence in practice. For non-conjugate likelihoods, we show that stochastic NGVI with the canonical parameterization implicitly optimizes a non-convex objective. Thus, a global convergence rate of $\mathcal{O}(\frac{1}{T})$ is unlikely without some significant new understanding of optimizing the ELBO using natural gradients. △ Less

Submitted 3 June, 2024; originally announced June 2024.

Comments: ICML 2024

Demystifying SGD with Doubly Stochastic Gradients

Authors: Kyurae Kim, Joohwan Ko, Yi-An Ma, Jacob R. Gardner

Abstract: Optimization objectives in the form of a sum of intractable expectations are rising in importance (e.g., diffusion models, variational autoencoders, and many more), a setting also known as "finite sum with infinite data." For these problems, a popular strategy is to employ SGD with doubly stochastic gradients (doubly SGD): the expectations are estimated using the gradient estimator of each compone… ▽ More Optimization objectives in the form of a sum of intractable expectations are rising in importance (e.g., diffusion models, variational autoencoders, and many more), a setting also known as "finite sum with infinite data." For these problems, a popular strategy is to employ SGD with doubly stochastic gradients (doubly SGD): the expectations are estimated using the gradient estimator of each component, while the sum is estimated by subsampling over these estimators. Despite its popularity, little is known about the convergence properties of doubly SGD, except under strong assumptions such as bounded variance. In this work, we establish the convergence of doubly SGD with independent minibatching and random reshuffling under general conditions, which encompasses dependent component gradient estimators. In particular, for dependent estimators, our analysis allows fined-grained analysis of the effect correlations. As a result, under a per-iteration computational budget of $b \times m$, where $b$ is the minibatch size and $m$ is the number of Monte Carlo samples, our analysis suggests where one should invest most of the budget in general. Furthermore, we prove that random reshuffling (RR) improves the complexity dependence on the subsampling noise. △ Less

Submitted 9 May, 2025; v1 submitted 2 June, 2024; originally announced June 2024.

Comments: Accepted to ICML'24; v2: fixed typo in complexity statements

Reimplementation of Learning to Reweight Examples for Robust Deep Learning

Authors: Parth Patil, Ben Boardley, Jack Gardner, Emily Loiselle, Deerajkumar Parthipan

Abstract: Deep neural networks (DNNs) have been used to create models for many complex analysis problems like image recognition and medical diagnosis. DNNs are a popular tool within machine learning due to their ability to model complex patterns and distributions. However, the performance of these networks is highly dependent on the quality of the data used to train the models. Two characteristics of these… ▽ More Deep neural networks (DNNs) have been used to create models for many complex analysis problems like image recognition and medical diagnosis. DNNs are a popular tool within machine learning due to their ability to model complex patterns and distributions. However, the performance of these networks is highly dependent on the quality of the data used to train the models. Two characteristics of these sets, noisy labels and training set biases, are known to frequently cause poor generalization performance as a result of overfitting to the training set. This paper aims to solve this problem using the approach proposed by Ren et al. (2018) using meta-training and online weight approximation. We will first implement a toy-problem to crudely verify the claims made by the authors of Ren et al. (2018) and then venture into using the approach to solve a real world problem of Skin-cancer detection using an imbalanced image dataset. △ Less

Submitted 10 May, 2024; originally announced May 2024.

SUNDIALS Time Integrators for Exascale Applications with Many Independent ODE Systems

Authors: Cody J. Balos, Marc Day, Lucas Esclapez, Anne M. Felden, David J. Gardner, Malik Hassanaly, Daniel R. Reynolds, Jon Rood, Jean M. Sexton, Nicholas T. Wimer, Carol S. Woodward

Abstract: Many complex systems can be accurately modeled as a set of coupled time-dependent partial differential equations (PDEs). However, solving such equations can be prohibitively expensive, easily taxing the world's largest supercomputers. One pragmatic strategy for attacking such problems is to split the PDEs into components that can more easily be solved in isolation. This operator splitting approach… ▽ More Many complex systems can be accurately modeled as a set of coupled time-dependent partial differential equations (PDEs). However, solving such equations can be prohibitively expensive, easily taxing the world's largest supercomputers. One pragmatic strategy for attacking such problems is to split the PDEs into components that can more easily be solved in isolation. This operator splitting approach is used ubiquitously across scientific domains, and in many cases leads to a set of ordinary differential equations (ODEs) that need to be solved as part of a larger "outer-loop" time-stepping approach. The SUNDIALS library provides a plethora of robust time integration algorithms for solving ODEs, and the U.S. Department of Energy Exascale Computing Project (ECP) has supported its extension to applications on exascale-capable computing hardware. In this paper, we highlight some SUNDIALS capabilities and its deployment in combustion and cosmology application codes (Pele and Nyx, respectively) where operator splitting gives rise to numerous, small ODE systems that must be solved concurrently. △ Less

Submitted 2 May, 2024; originally announced May 2024.

Stochastic Approximation with Biased MCMC for Expectation Maximization

Authors: Samuel Gruffaz, Kyurae Kim, Alain Oliviero Durmus, Jacob R. Gardner

Abstract: The expectation maximization (EM) algorithm is a widespread method for empirical Bayesian inference, but its expectation step (E-step) is often intractable. Employing a stochastic approximation scheme with Markov chain Monte Carlo (MCMC) can circumvent this issue, resulting in an algorithm known as MCMC-SAEM. While theoretical guarantees for MCMC-SAEM have previously been established, these result… ▽ More The expectation maximization (EM) algorithm is a widespread method for empirical Bayesian inference, but its expectation step (E-step) is often intractable. Employing a stochastic approximation scheme with Markov chain Monte Carlo (MCMC) can circumvent this issue, resulting in an algorithm known as MCMC-SAEM. While theoretical guarantees for MCMC-SAEM have previously been established, these results are restricted to the case where asymptotically unbiased MCMC algorithms are used. In practice, MCMC-SAEM is often run with asymptotically biased MCMC, for which the consequences are theoretically less understood. In this work, we fill this gap by analyzing the asymptotics and non-asymptotics of SAEM with biased MCMC steps, particularly the effect of bias. We also provide numerical experiments comparing the Metropolis-adjusted Langevin algorithm (MALA), which is asymptotically unbiased, and the unadjusted Langevin algorithm (ULA), which is asymptotically biased, on synthetic and real datasets. Experimental results show that ULA is more stable with respect to the choice of Langevin stepsize and can sometimes result in faster convergence. △ Less

Submitted 27 February, 2024; originally announced February 2024.

Comments: Accepted to AISTATS'24

Generative Adversarial Model-Based Optimization via Source Critic Regularization

Authors: Michael S. Yao, Yimeng Zeng, Hamsa Bastani, Jacob Gardner, James C. Gee, Osbert Bastani

Abstract: Offline model-based optimization seeks to optimize against a learned surrogate model without querying the true oracle objective function during optimization. Such tasks are commonly encountered in protein design, robotics, and clinical medicine where evaluating the oracle function is prohibitively expensive. However, inaccurate surrogate model predictions are frequently encountered along offline o… ▽ More Offline model-based optimization seeks to optimize against a learned surrogate model without querying the true oracle objective function during optimization. Such tasks are commonly encountered in protein design, robotics, and clinical medicine where evaluating the oracle function is prohibitively expensive. However, inaccurate surrogate model predictions are frequently encountered along offline optimization trajectories. To address this limitation, we propose generative adversarial model-based optimization using adaptive source critic regularization (aSCR) -- a task- and optimizer- agnostic framework for constraining the optimization trajectory to regions of the design space where the surrogate function is reliable. We propose a computationally tractable algorithm to dynamically adjust the strength of this constraint, and show how leveraging aSCR with standard Bayesian optimization outperforms existing methods on a suite of offline generative design tasks. Our code is available at https://github.com/michael-s-yao/gabo △ Less

Submitted 25 September, 2024; v1 submitted 9 February, 2024; originally announced February 2024.

Comments: 31 pages, Accepted to NeurIPS 2024

Provably Scalable Black-Box Variational Inference with Structured Variational Families

Authors: Joohwan Ko, Kyurae Kim, Woo Chang Kim, Jacob R. Gardner

Abstract: Variational families with full-rank covariance approximations are known not to work well in black-box variational inference (BBVI), both empirically and theoretically. In fact, recent computational complexity results for BBVI have established that full-rank variational families scale poorly with the dimensionality of the problem compared to e.g. mean-field families. This is particularly critical t… ▽ More Variational families with full-rank covariance approximations are known not to work well in black-box variational inference (BBVI), both empirically and theoretically. In fact, recent computational complexity results for BBVI have established that full-rank variational families scale poorly with the dimensionality of the problem compared to e.g. mean-field families. This is particularly critical to hierarchical Bayesian models with local variables; their dimensionality increases with the size of the datasets. Consequently, one gets an iteration complexity with an explicit $\mathcal{O}(N^2)$ dependence on the dataset size $N$. In this paper, we explore a theoretical middle ground between mean-field variational families and full-rank families: structured variational families. We rigorously prove that certain scale matrix structures can achieve a better iteration complexity of $\mathcal{O}\left(N\right)$, implying better scaling with respect to $N$. We empirically verify our theoretical results on large-scale hierarchical models. △ Less

Submitted 13 November, 2025; v1 submitted 19 January, 2024; originally announced January 2024.

Comments: Accepted to ICML'24; v3, v4: fixed typos

Distributional Latent Variable Models with an Application in Active Cognitive Testing

Authors: Robert Kasumba, Dom CP Marticorena, Anja Pahor, Geetha Ramani, Imani Goffney, Susanne M Jaeggi, Aaron Seitz, Jacob R Gardner, Dennis L Barbour

Abstract: Cognitive modeling commonly relies on asking participants to complete a battery of varied tests in order to estimate attention, working memory, and other latent variables. In many cases, these tests result in highly variable observation models. A near-ubiquitous approach is to repeat many observations for each test independently, resulting in a distribution over the outcomes from each test given t… ▽ More Cognitive modeling commonly relies on asking participants to complete a battery of varied tests in order to estimate attention, working memory, and other latent variables. In many cases, these tests result in highly variable observation models. A near-ubiquitous approach is to repeat many observations for each test independently, resulting in a distribution over the outcomes from each test given to each subject. Latent variable models (LVMs), if employed, are only added after data collection. In this paper, we explore the usage of LVMs to enable learning across many correlated variables simultaneously. We extend LVMs to the setting where observed data for each subject are a series of observations from many different distributions, rather than simple vectors to be reconstructed. By embedding test battery results for individuals in a latent space that is trained jointly across a population, we can leverage correlations both between disparate test data for a single participant and between multiple participants. We then propose an active learning framework that leverages this model to conduct more efficient cognitive test batteries. We validate our approach by demonstrating with real-time data acquisition that it performs comparably to conventional methods in making item-level predictions with fewer test items. △ Less

Submitted 25 September, 2024; v1 submitted 14 December, 2023; originally announced December 2023.

Comments: 11 pages, 6 figures

Benchmarking Distribution Shift in Tabular Data with TableShift

Authors: Josh Gardner, Zoran Popovic, Ludwig Schmidt

Abstract: Robustness to distribution shift has become a growing concern for text and image models as they transition from research subjects to deployment in the real world. However, high-quality benchmarks for distribution shift in tabular machine learning tasks are still lacking despite the widespread real-world use of tabular data and differences in the models used for tabular data in comparison to text a… ▽ More Robustness to distribution shift has become a growing concern for text and image models as they transition from research subjects to deployment in the real world. However, high-quality benchmarks for distribution shift in tabular machine learning tasks are still lacking despite the widespread real-world use of tabular data and differences in the models used for tabular data in comparison to text and images. As a consequence, the robustness of tabular models to distribution shift is poorly understood. To address this issue, we introduce TableShift, a distribution shift benchmark for tabular data. TableShift contains 15 binary classification tasks in total, each with an associated shift, and includes a diverse set of data sources, prediction targets, and distribution shifts. The benchmark covers domains including finance, education, public policy, healthcare, and civic participation, and is accessible using only a few lines of Python code via the TableShift API. We conduct a large-scale study comparing several state-of-the-art tabular data models alongside robust learning and domain generalization methods on the benchmark tasks. Our study demonstrates (1) a linear trend between in-distribution (ID) and out-of-distribution (OOD) accuracy; (2) domain robustness methods can reduce shift gaps but at the cost of reduced ID accuracy; (3) a strong relationship between shift gap (difference between ID and OOD performance) and shifts in the label distribution. The benchmark data, Python package, model implementations, and more information about TableShift are available at https://github.com/mlfoundations/tableshift and https://tableshift.org . △ Less

Submitted 8 February, 2024; v1 submitted 10 December, 2023; originally announced December 2023.

Comments: NeurIPS 2023 Dataset and Benchmarks Track accepted version

The Sky's the Limit: Re-lightable Outdoor Scenes via a Sky-pixel Constrained Illumination Prior and Outside-In Visibility

Authors: James A. D. Gardner, Evgenii Kashin, Bernhard Egger, William A. P. Smith

Abstract: Inverse rendering of outdoor scenes from unconstrained image collections is a challenging task, particularly illumination/albedo ambiguities and occlusion of the illumination environment (shadowing) caused by geometry. However, there are many cues in an image that can aid in the disentanglement of geometry, albedo and shadows. Whilst sky is frequently masked out in state-of-the-art methods, we exp… ▽ More Inverse rendering of outdoor scenes from unconstrained image collections is a challenging task, particularly illumination/albedo ambiguities and occlusion of the illumination environment (shadowing) caused by geometry. However, there are many cues in an image that can aid in the disentanglement of geometry, albedo and shadows. Whilst sky is frequently masked out in state-of-the-art methods, we exploit the fact that any sky pixel provides a direct observation of distant lighting in the corresponding direction and, via a neural illumination prior, a statistical cue to derive the remaining illumination environment. The incorporation of our illumination prior is enabled by a novel `outside-in' method for computing differentiable sky visibility based on a neural directional distance function. This is highly efficient and can be trained in parallel with the neural scene representation, allowing gradients from appearance loss to flow from shadows to influence the estimation of illumination and geometry. Our method estimates high-quality albedo, geometry, illumination and sky visibility, achieving state-of-the-art results on the NeRF-OSR relighting benchmark. Our code and models can be found at https://github.com/JADGardner/neusky △ Less

Submitted 30 July, 2024; v1 submitted 28 November, 2023; originally announced November 2023.

Comments: Accepted to ECCV 2024

RENI++ A Rotation-Equivariant, Scale-Invariant, Natural Illumination Prior

Authors: James A. D. Gardner, Bernhard Egger, William A. P. Smith

Abstract: Inverse rendering is an ill-posed problem. Previous work has sought to resolve this by focussing on priors for object or scene shape or appearance. In this work, we instead focus on a prior for natural illuminations. Current methods rely on spherical harmonic lighting or other generic representations and, at best, a simplistic prior on the parameters. This results in limitations for the inverse se… ▽ More Inverse rendering is an ill-posed problem. Previous work has sought to resolve this by focussing on priors for object or scene shape or appearance. In this work, we instead focus on a prior for natural illuminations. Current methods rely on spherical harmonic lighting or other generic representations and, at best, a simplistic prior on the parameters. This results in limitations for the inverse setting in terms of the expressivity of the illumination conditions, especially when taking specular reflections into account. We propose a conditional neural field representation based on a variational auto-decoder and a transformer decoder. We extend Vector Neurons to build equivariance directly into our architecture, and leveraging insights from depth estimation through a scale-invariant loss function, we enable the accurate representation of High Dynamic Range (HDR) images. The result is a compact, rotation-equivariant HDR neural illumination model capable of capturing complex, high-frequency features in natural environment maps. Training our model on a curated dataset of 1.6K HDR environment maps of natural scenes, we compare it against traditional representations, demonstrate its applicability for an inverse rendering task and show environment map completion from partial observations. We share our PyTorch implementation, dataset and trained models at https://github.com/JADGardner/ns_reni △ Less

Submitted 15 November, 2023; originally announced November 2023.

Comments: Project Repo - https://github.com/JADGardner/ns_reni. arXiv admin note: substantial text overlap with arXiv:2206.03858

Large-Scale Gaussian Processes via Alternating Projection

Authors: Kaiwen Wu, Jonathan Wenger, Haydn Jones, Geoff Pleiss, Jacob R. Gardner

Abstract: Training and inference in Gaussian processes (GPs) require solving linear systems with $n\times n$ kernel matrices. To address the prohibitive $\mathcal{O}(n^3)$ time complexity, recent work has employed fast iterative methods, like conjugate gradients (CG). However, as datasets increase in magnitude, the kernel matrices become increasingly ill-conditioned and still require $\mathcal{O}(n^2)$ spac… ▽ More Training and inference in Gaussian processes (GPs) require solving linear systems with $n\times n$ kernel matrices. To address the prohibitive $\mathcal{O}(n^3)$ time complexity, recent work has employed fast iterative methods, like conjugate gradients (CG). However, as datasets increase in magnitude, the kernel matrices become increasingly ill-conditioned and still require $\mathcal{O}(n^2)$ space without partitioning. Thus, while CG increases the size of datasets GPs can be trained on, modern datasets reach scales beyond its applicability. In this work, we propose an iterative method which only accesses subblocks of the kernel matrix, effectively enabling mini-batching. Our algorithm, based on alternating projection, has $\mathcal{O}(n)$ per-iteration time and space complexity, solving many of the practical challenges of scaling GPs to very large datasets. Theoretically, we prove the method enjoys linear convergence. Empirically, we demonstrate its fast convergence in practice and robustness to ill-conditioning. On large-scale benchmark datasets with up to four million data points, our approach accelerates GP training and inference by speed-up factors up to $27\times$ and $72 \times$, respectively, compared to CG. △ Less

Submitted 8 March, 2024; v1 submitted 26 October, 2023; originally announced October 2023.

Comments: AISTATS 2024

LLark: A Multimodal Instruction-Following Language Model for Music

Authors: Josh Gardner, Simon Durand, Daniel Stoller, Rachel M. Bittner

Abstract: Music has a unique and complex structure which is challenging for both expert humans and existing AI systems to understand, and presents unique challenges relative to other forms of audio. We present LLark, an instruction-tuned multimodal model for \emph{music} understanding. We detail our process for dataset creation, which involves augmenting the annotations of diverse open-source music datasets… ▽ More Music has a unique and complex structure which is challenging for both expert humans and existing AI systems to understand, and presents unique challenges relative to other forms of audio. We present LLark, an instruction-tuned multimodal model for \emph{music} understanding. We detail our process for dataset creation, which involves augmenting the annotations of diverse open-source music datasets and converting them to a unified instruction-tuning format. We propose a multimodal architecture for LLark, integrating a pretrained generative model for music with a pretrained language model. In evaluations on three types of tasks (music understanding, captioning, reasoning), we show that LLark matches or outperforms existing baselines in music understanding, and that humans show a high degree of agreement with its responses in captioning and reasoning tasks. LLark is trained entirely from open-source music data and models, and we make our training code available along with the release of this paper. Additional results and audio examples are at https://bit.ly/llark, and our source code is available at https://github.com/spotify-research/llark . △ Less

Submitted 2 June, 2024; v1 submitted 10 October, 2023; originally announced October 2023.

Comments: ICML camera-ready version

Optimal Stopping with Multi-Dimensional Comparative Loss Aversion

Authors: Linda Cai, Joshua Gardner, S. Matthew Weinberg

Abstract: Despite having the same basic prophet inequality setup and model of loss aversion, conclusions in our multi-dimensional model differs considerably from the one-dimensional model of Kleinberg et al. For example, Kleinberg et al. gives a tight closed-form on the competitive ratio that an online decision-maker can achieve as a function of $λ$, for any $λ\geq 0$. In our multi-dimensional model, there… ▽ More Despite having the same basic prophet inequality setup and model of loss aversion, conclusions in our multi-dimensional model differs considerably from the one-dimensional model of Kleinberg et al. For example, Kleinberg et al. gives a tight closed-form on the competitive ratio that an online decision-maker can achieve as a function of $λ$, for any $λ\geq 0$. In our multi-dimensional model, there is a sharp phase transition: if $k$ denotes the number of dimensions, then when $λ\cdot (k-1) \geq 1$, no non-trivial competitive ratio is possible. On the other hand, when $λ\cdot (k-1) < 1$, we give a tight bound on the achievable competitive ratio (similar to Kleinberg et al.). As another example, Kleinberg et al. uncovers an exponential improvement in their competitive ratio for the random-order vs. worst-case prophet inequality problem. In our model with $k\geq 2$ dimensions, the gap is at most a constant-factor. We uncover several additional key differences in the multi- and single-dimensional models. △ Less

Submitted 26 September, 2023; v1 submitted 25 September, 2023; originally announced September 2023.

Comments: Accepted to WINE 2023

VisIT-Bench: A Benchmark for Vision-Language Instruction Following Inspired by Real-World Use

Authors: Yonatan Bitton, Hritik Bansal, Jack Hessel, Rulin Shao, Wanrong Zhu, Anas Awadalla, Josh Gardner, Rohan Taori, Ludwig Schmidt

Abstract: We introduce VisIT-Bench (Visual InsTruction Benchmark), a benchmark for evaluation of instruction-following vision-language models for real-world use. Our starting point is curating 70 'instruction families' that we envision instruction tuned vision-language models should be able to address. Extending beyond evaluations like VQAv2 and COCO, tasks range from basic recognition to game playing and c… ▽ More We introduce VisIT-Bench (Visual InsTruction Benchmark), a benchmark for evaluation of instruction-following vision-language models for real-world use. Our starting point is curating 70 'instruction families' that we envision instruction tuned vision-language models should be able to address. Extending beyond evaluations like VQAv2 and COCO, tasks range from basic recognition to game playing and creative generation. Following curation, our dataset comprises 592 test queries, each with a human-authored instruction-conditioned caption. These descriptions surface instruction-specific factors, e.g., for an instruction asking about the accessibility of a storefront for wheelchair users, the instruction-conditioned caption describes ramps/potential obstacles. These descriptions enable 1) collecting human-verified reference outputs for each instance; and 2) automatic evaluation of candidate multimodal generations using a text-only LLM, aligning with human judgment. We quantify quality gaps between models and references using both human and automatic evaluations; e.g., the top-performing instruction-following model wins against the GPT-4 reference in just 27% of the comparison. VisIT-Bench is dynamic to participate, practitioners simply submit their model's response on the project website; Data, code and leaderboard is available at visit-bench.github.io. △ Less

Submitted 26 December, 2023; v1 submitted 12 August, 2023; originally announced August 2023.

Comments: Accepted to NeurIPS 2023, Datasets and Benchmarks. Website: https://visit-bench.github.io/

OpenFlamingo: An Open-Source Framework for Training Large Autoregressive Vision-Language Models

Authors: Anas Awadalla, Irena Gao, Josh Gardner, Jack Hessel, Yusuf Hanafy, Wanrong Zhu, Kalyani Marathe, Yonatan Bitton, Samir Gadre, Shiori Sagawa, Jenia Jitsev, Simon Kornblith, Pang Wei Koh, Gabriel Ilharco, Mitchell Wortsman, Ludwig Schmidt

Abstract: We introduce OpenFlamingo, a family of autoregressive vision-language models ranging from 3B to 9B parameters. OpenFlamingo is an ongoing effort to produce an open-source replication of DeepMind's Flamingo models. On seven vision-language datasets, OpenFlamingo models average between 80 - 89% of corresponding Flamingo performance. This technical report describes our models, training data, hyperpar… ▽ More We introduce OpenFlamingo, a family of autoregressive vision-language models ranging from 3B to 9B parameters. OpenFlamingo is an ongoing effort to produce an open-source replication of DeepMind's Flamingo models. On seven vision-language datasets, OpenFlamingo models average between 80 - 89% of corresponding Flamingo performance. This technical report describes our models, training data, hyperparameters, and evaluation suite. We share our models and code at https://github.com/mlfoundations/open_flamingo. △ Less

Submitted 7 August, 2023; v1 submitted 2 August, 2023; originally announced August 2023.