arXiv subjects

Anthony Chen

Publications and source records attributed to Anthony Chen.

30 records · Page 2Linked to original sources

arXiv2024

A Baseline Analysis of Reward Models' Ability To Accurately Analyze Foundation Models Under Distribution Shift

Foundation models, specifically Large Language Models (LLMs), have lately gained wide-spread attention and adoption. Reinforcement Learning with Human Feedback (RLHF) involves training a reward model to capture desired behaviors, which is then used to align LLM's. These reward models are additionally used at inference-time to estimate LLM responses' adherence to those desired behaviors. However, there is little work measuring how robust these reward models are to distribution shifts. In this work, we evaluate how reward model performance - measured via accuracy and calibration (i.e. alignment between accuracy and confidence) - is affected by distribution shift. We show novel calibration patterns and accuracy drops due to OOD prompts and responses, and that the reward model is more sensitive to shifts in responses than prompts. Additionally, we adapt an OOD detection technique commonly used in classification to the reward model setting to detect these distribution shifts in prompts and responses.

cs.CL↗

arXiv2024

Fast Summation on the Sphere with Applications to the Barotropic Vorticity Equation

Fast summation refers to a family of techniques for approximating $O(N^2)$ sums in $O(N\log{N})$ or $O(N)$ time. These techniques have traditionally found wide use in astrophysics and electrostatics in calculating the forces in a $N$-body problem. In this work, we present a spherical tree code, and apply it to the problem of efficiently solving the barotropic vorticity equation.

math.NA↗

arXiv2023

The Data Provenance Initiative: A Large Scale Audit of Dataset Licensing & Attribution in AI

The race to train language models on vast, diverse, and inconsistently documented datasets has raised pressing concerns about the legal and ethical risks for practitioners. To remedy these practices threatening data transparency and understanding, we convene a multi-disciplinary effort between legal and machine learning experts to systematically audit and trace 1800+ text datasets. We develop tools and standards to trace the lineage of these datasets, from their source, creators, series of license conditions, properties, and subsequent use. Our landscape analysis highlights the sharp divides in composition and focus of commercially open vs closed datasets, with closed datasets monopolizing important categories: lower resource languages, more creative tasks, richer topic variety, newer and more synthetic training data. This points to a deepening divide in the types of data that are made available under different license conditions, and heightened implications for jurisdictional legal interpretations of copyright and fair use. We also observe frequent miscategorization of licenses on widely used dataset hosting sites, with license omission of 70%+ and error rates of 50%+. This points to a crisis in misattribution and informed use of the most popular datasets driving many recent breakthroughs. As a contribution to ongoing improvements in dataset transparency and responsible use, we release our entire audit, with an interactive UI, the Data Provenance Explorer, which allows practitioners to trace and filter on data provenance for the most popular open source finetuning data collections: www.dataprovenance.org.

cs.CL↗

arXiv2023

RARR: Researching and Revising What Language Models Say, Using Language Models

Language models (LMs) now excel at many tasks such as few-shot learning, question answering, reasoning, and dialog. However, they sometimes generate unsupported or misleading content. A user cannot easily determine whether their outputs are trustworthy or not, because most LMs do not have any built-in mechanism for attribution to external evidence. To enable attribution while still preserving all the powerful advantages of recent generation models, we propose RARR (Retrofit Attribution using Research and Revision), a system that 1) automatically finds attribution for the output of any text generation model and 2) post-edits the output to fix unsupported content while preserving the original output as much as possible. When applied to the output of several state-of-the-art LMs on a diverse set of generation tasks, we find that RARR significantly improves attribution while otherwise preserving the original input to a much greater degree than previously explored edit models. Furthermore, the implementation of RARR requires only a handful of training examples, a large language model, and standard web search.

cs.CL↗

arXiv2023

PURR: Efficiently Editing Language Model Hallucinations by Denoising Language Model Corruptions

The remarkable capabilities of large language models have been accompanied by a persistent drawback: the generation of false and unsubstantiated claims commonly known as "hallucinations". To combat this issue, recent research has introduced approaches that involve editing and attributing the outputs of language models, particularly through prompt-based editing. However, the inference cost and speed of using large language models for editing currently bottleneck prompt-based methods. These bottlenecks motivate the training of compact editors, which is challenging due to the scarcity of training data for this purpose. To overcome these challenges, we exploit the power of large language models to introduce corruptions (i.e., noise) into text and subsequently fine-tune compact editors to denoise the corruptions by incorporating relevant evidence. Our methodology is entirely unsupervised and provides us with faux hallucinations for training in any domain. Our Petite Unsupervised Research and Revision model, PURR, not only improves attribution over existing editing methods based on fine-tuning and prompting, but also achieves faster execution times by orders of magnitude.

cs.CL↗

arXiv2023

PiMAE: Point Cloud and Image Interactive Masked Autoencoders for 3D Object Detection

Masked Autoencoders learn strong visual representations and achieve state-of-the-art results in several independent modalities, yet very few works have addressed their capabilities in multi-modality settings. In this work, we focus on point cloud and RGB image data, two modalities that are often presented together in the real world, and explore their meaningful interactions. To improve upon the cross-modal synergy in existing works, we propose PiMAE, a self-supervised pre-training framework that promotes 3D and 2D interaction through three aspects. Specifically, we first notice the importance of masking strategies between the two sources and utilize a projection module to complementarily align the mask and visible tokens of the two modalities. Then, we utilize a well-crafted two-branch MAE pipeline with a novel shared decoder to promote cross-modality interaction in the mask tokens. Finally, we design a unique cross-modal reconstruction module to enhance representation learning for both modalities. Through extensive experiments performed on large-scale RGB-D scene understanding benchmarks (SUN RGB-D and ScannetV2), we discover it is nontrivial to interactively learn point-image features, where we greatly improve multiple 3D detectors, 2D detectors, and few-shot classifiers by 2.9%, 6.7%, and 2.4%, respectively. Code is available at https://github.com/BLVLab/PiMAE.

cs.CV↗

arXiv2022

Travel time reliability in transportation networks: A review of methodological developments

The unavoidable travel time variability in transportation networks, resulted from the widespread supply side and demand side uncertainties, makes travel time reliability (TTR) be a common and core interest of all the stakeholders in transportation systems, including planners, travelers, service providers, and managers. This common and core interest stimulates extensive studies on modeling TTR. Researchers have developed a range of theories and models of TTR, many of which have been incorporated into transportation models, policies, and project appraisals. Adopting the network perspective, this paper aims to provide an integrated framework for reviewing the methodological developments of modeling TTR in transportation networks, including its characterization, evaluation and valuation, and traffic assignment. Specifically, the TTR characterization provides a whole picture of travel time distribution in transportation networks. TTR evaluation and TTR valuation (known as the value of reliability, VOR) simply and intuitively interpret abstract characterized TTR to be well understood by different stakeholders of transportation systems. TTR-based traffic assignment investigates the effects of TTR on the individual users travel behavior and consequently the collective network flow pattern. As the above three topics are mainly separately studied in different disciplines and research areas, the integrated framework allows us to better understand their relationships and may contribute to developing possible combinations of TTR modeling philosophy. Also, the network perspective enables to focus on common challenges of modeling TTR, especially the uncertainty propagation from the uncertainty sources to the TTR at spatial levels including link, route, and the entire network. Some directions for future research are discussed in the era of new data environment, applications, and emerging technologies.

econ.GN↗

arXiv2022

Communication Bounds for Convolutional Neural Networks

Convolutional neural networks (CNNs) are important in a wide variety of machine learning tasks and applications, so optimizing their performance is essential. Moving words of data between levels of a memory hierarchy or between processors on a network is much more expensive than the cost of arithmetic, so minimizing communication is critical to optimizing performance. In this paper, we present new lower bounds on data movement for mixed precision convolutions in both single-processor and parallel distributed memory models, as well as algorithms that outperform current implementations such as Im2Col. We obtain performance figures using GEMMINI, a machine learning accelerator, where our tiling provides improvements between 13% and 150% over a vendor supplied algorithm.

cs.DC↗

arXiv2022

Entity-Based Knowledge Conflicts in Question Answering

Knowledge-dependent tasks typically use two sources of knowledge: parametric, learned at training time, and contextual, given as a passage at inference time. To understand how models use these sources together, we formalize the problem of knowledge conflicts, where the contextual information contradicts the learned information. Analyzing the behaviour of popular models, we measure their over-reliance on memorized information (the cause of hallucinations), and uncover important factors that exacerbate this behaviour. Lastly, we propose a simple method to mitigate over-reliance on parametric knowledge, which minimizes hallucination, and improves out-of-distribution generalization by 4%-7%. Our findings demonstrate the importance for practitioners to evaluate model tendency to hallucinate rather than read, and show that our mitigation strategy encourages generalization to evolving information (i.e., time-dependent queries). To encourage these practices, we have released our framework for generating knowledge conflicts.

cs.CL↗

arXiv2021

Evaluating Entity Disambiguation and the Role of Popularity in Retrieval-Based NLP

Retrieval is a core component for open-domain NLP tasks. In open-domain tasks, multiple entities can share a name, making disambiguation an inherent yet under-explored problem. We propose an evaluation benchmark for assessing the entity disambiguation capabilities of these retrievers, which we call Ambiguous Entity Retrieval (AmbER) sets. We define an AmbER set as a collection of entities that share a name along with queries about those entities. By covering the set of entities for polysemous names, AmbER sets act as a challenging test of entity disambiguation. We create AmbER sets for three popular open-domain tasks: fact checking, slot filling, and question answering, and evaluate a diverse set of retrievers. We find that the retrievers exhibit popularity bias, significantly under-performing on rarer entities that share a name, e.g., they are twice as likely to retrieve erroneous documents on queries for the less popular entity under the same name. These experiments on AmbER sets show their utility as an evaluation tool and highlight the weaknesses of popular retrieval systems.

cs.CL↗

arXiv2020

MOCHA: A Dataset for Training and Evaluating Generative Reading Comprehension Metrics

Posing reading comprehension as a generation problem provides a great deal of flexibility, allowing for open-ended questions with few restrictions on possible answers. However, progress is impeded by existing generation metrics, which rely on token overlap and are agnostic to the nuances of reading comprehension. To address this, we introduce a benchmark for training and evaluating generative reading comprehension metrics: MOdeling Correctness with Human Annotations. MOCHA contains 40K human judgement scores on model outputs from 6 diverse question answering datasets and an additional set of minimal pairs for evaluation. Using MOCHA, we train a Learned Evaluation metric for Reading Comprehension, LERC, to mimic human judgement scores. LERC outperforms baseline metrics by 10 to 36 absolute Pearson points on held-out annotations. When we evaluate robustness on minimal pairs, LERC achieves 80% accuracy, outperforming baselines by 14 to 26 absolute percentage points while leaving significant room for improvement. MOCHA presents a challenging problem for developing accurate and robust generative reading comprehension metrics.

cs.CL↗

arXiv2019

Maximizing Energy Battery Efficiency in Swarm Robotics

Miniaturization and cost, two of the main attractive factors of swarm robotics, have motivated its use as a solution in object collecting tasks, search & rescue missions, and other applications. However, in the current literature only a few papers consider energy allocation efficiency within a swarm. Generally, robots recharge to their maximum level every time unconditionally, and do not incorporate estimates of the energy needed for their next task. In this paper we present an energy efficiency maximization method that minimizes the overall energy cost within a swarm while simultaneously maximizing swarm performance on an object gathering task. The method utilizes dynamic thresholds for upper and lower battery limits. This method has also shown to improve the efficiency of existing energy management methods.

cs.MA↗