Search

Large Language Models (LLMs) have shown impressive performance in complex reasoning tasks through Chain-of-Thought (CoT) reasoning, allowing models to break down problems into manageable sub-tasks. However, existing CoT evaluation techniques either require annotated CoT data or fall short in accurately assessing intermediate reasoning steps, leading to high rates of false positives. In this paper, we formalize CoT reasoning in LLMs through an information-theoretic lens. Specifically, our framework quantifies the `information gain' at each reasoning step, enabling the identification of failure modes in LLMs without the need for expensive annotated datasets. We demonstrate the efficacy of our approach through extensive experiments on toy and GSM-8K data, where it significantly outperforms existing outcome-based methods by providing more accurate insights into model performance on individual tasks.

The Bradley-Terry (BT) model is a common and successful practice in reward modeling for Large Language Model (LLM) alignment. However, it remains unclear why this model -- originally developed for multi-player stochastic game matching -- can be adopted to convert pairwise response comparisons to reward values and make predictions. Especially given the fact that only a limited number of prompt-response pairs are sparsely compared with others. In this paper, we first revisit the foundations of using BT models in reward modeling, and establish the convergence rate of BT reward models based on deep neural networks using embeddings, providing a theoretical foundation for their use. Despite theoretically sound, we argue that the BT model is not a necessary choice from the perspective of downstream optimization. This is because a reward model only needs to preserve the correct ranking predictions through a monotonic transformation of the true reward. We highlight the critical concept of order consistency in reward modeling and demonstrate that the BT model possesses this property. Consequently, we propose a simple and straightforward upper-bound algorithm, compatible with off-the-shelf binary classifiers, as an alternative order-consistent reward modeling objective. To offer practical insights, we empirically evaluate the performance of these different reward modeling approaches across more than 12,000 experimental setups, using $6$ base LLMs, $2$ datasets, and diverse annotation designs that vary in quantity, quality, and pairing choices in preference annotations.

Direct Z-scheme heterostructures with enhanced redox potential are increasingly regarded as promising materials for solar-driven water splitting. This potential arises from the synergistic interaction between the intrinsic dipoles in Janus materials and the interfacial electric fields across the layers. In this study, we explore the photocatalytic potential of 20 two-dimensional (2D) Janus transition metal dichalcogenide (TMDC) heterobilayers for efficient water splitting. Utilizing density functional theory (DFT) calculations, we first screen these materials based on key properties such as band gaps and the magnitude of intrinsic electric fields to identify promising candidates. We then evaluate additional critical factors, including carrier mobility and surface chemical reactions, to fully assess their performance. The intrinsic dipole moments in Janus materials generate built-in electric fields that enhance charge separation and reduce carrier recombination, thereby improving photocatalytic efficiency. Furthermore, we employ the Fr\"{o}hlich interaction model to quantify the mobility contributions from the longitudinal optical phonon mode, providing detailed insights into how carrier mobility, influenced by phonon scattering, affects photocatalytic performance. Our results reveal that several Janus-TMDC heterobilayers, including WSe$_2$-SWSe, WSe$_2$-TeWSe, and WS$_2$-SMoSe, exhibit strong absorption in the visible spectrum and achieve solar-to-hydrogen (STH) conversion efficiencies of up to 33.24%. These findings demonstrate the potential of Janus-based Z-scheme systems to overcome existing limitations in photocatalytic water splitting by optimizing the electronic and structural properties of 2D materials. This research highlights a viable pathway for advancing clean energy generation through enhanced photocatalytic processes., Comment: 30 pages, 6 figures, 4 tables

Large language models (LLMs) have demonstrated a remarkable ability to serve as general-purpose tools for various language-based tasks. Recent works have demonstrated that the efficacy of such models can be improved through iterative dialog between multiple models, frequently referred to as multi-agent debate (MAD). While debate shows promise as a means of improving model efficacy, most works in this area treat debate as an emergent behavior, rather than a learned behavior. In doing so, current debate frameworks rely on collaborative behaviors to have been sufficiently trained into off-the-shelf models. To address this limitation, we propose ACC-Debate, an Actor-Critic based learning framework to produce a two-agent team specialized in debate. We demonstrate that ACC-Debate outperforms SotA debate techniques on a wide array of benchmarks.

Probing spatially confined quantum states from afar - a long-sought goal to minimize external interference - has been proposed to be achievable in condensed matter systems via coherent projection. The latter can be tailored by sculpturing the eigenstates of the electron sea that surrounds the quantum state using atom-by-atom built cages, so-called quantum corrals. However, assuring the coherent nature of the projection, and manipulating its quantum composition, has remained an elusive goal. Here, we experimentally realize the coherent projection of a magnetic impurity-induced, Yu-Shiba-Rusinov quantum state using the eigenmodes of corrals on the surface of a superconductor, which enables us to manipulate the particle-hole composition of the projected state by tuning corral eigenmodes through the Fermi energy. Our results demonstrate a controlled non-local method for the detection of magnet superconductor hybrid quantum states., Comment: 19 pages, 5 figures

We introduce MDSGen, a novel framework for vision-guided open-domain sound generation optimized for model parameter size, memory consumption, and inference speed. This framework incorporates two key innovations: (1) a redundant video feature removal module that filters out unnecessary visual information, and (2) a temporal-aware masking strategy that leverages temporal context for enhanced audio generation accuracy. In contrast to existing resource-heavy Unet-based models, MDSGen employs denoising masked diffusion transformers, facilitating efficient generation without reliance on pre-trained diffusion models. Evaluated on the benchmark VGGSound dataset, our smallest model (5M parameters) achieves 97.9% alignment accuracy, using 172x fewer parameters, 371% less memory, and offering 36x faster inference than the current 860M-parameter state-of-the-art model (93.9% accuracy). The larger model (131M parameters) reaches nearly 99% accuracy while requiring 6.5x fewer parameters. These results highlight the scalability and effectiveness of our approach., Comment: 21 pages, 16 figures

This paper introduces SGCode, a flexible prompt-optimizing system to generate secure code with large language models (LLMs). SGCode integrates recent prompt-optimization approaches with LLMs in a unified system accessible through front-end and back-end APIs, enabling users to 1) generate secure code, which is free of vulnerabilities, 2) review and share security analysis, and 3) easily switch from one prompt optimization approach to another, while providing insights on model and system performance. We populated SGCode on an AWS server with PromSec, an approach that optimizes prompts by combining an LLM and security tools with a lightweight generative adversarial graph neural network to detect and fix security vulnerabilities in the generated code. Extensive experiments show that SGCode is practical as a public tool to gain insights into the trade-offs between model utility, secure code generation, and system cost. SGCode has only a marginal cost compared with prompting LLMs. SGCode is available at: https://sgcode.codes/.

Camera trap imagery has become an invaluable asset in contemporary wildlife surveillance, enabling researchers to observe and investigate the behaviors of wild animals. While existing methods rely solely on image data for classification, this may not suffice in cases of suboptimal animal angles, lighting, or image quality. This study introduces a novel approach that enhances wild animal classification by combining specific metadata (temperature, location, time, etc) with image data. Using a dataset focused on the Norwegian climate, our models show an accuracy increase from 98.4% to 98.9% compared to existing methods. Notably, our approach also achieves high accuracy with metadata-only classification, highlighting its potential to reduce reliance on image quality. This work paves the way for integrated systems that advance wildlife classification technology.

Open-vocabulary 3D instance segmentation transcends traditional closed-vocabulary methods by enabling the identification of both previously seen and unseen objects in real-world scenarios. It leverages a dual-modality approach, utilizing both 3D point clouds and 2D multi-view images to generate class-agnostic object mask proposals. Previous efforts predominantly focused on enhancing 3D mask proposal models; consequently, the information that could come from 2D association to 3D was not fully exploited. This bias towards 3D data, while effective for familiar indoor objects, limits the system's adaptability to new and varied object types, where 2D models offer greater utility. Addressing this gap, we introduce Zero-Shot Dual-Path Integration Framework that equally values the contributions of both 3D and 2D modalities. Our framework comprises three components: 3D pathway, 2D pathway, and Dual-Path Integration. 3D pathway generates spatially accurate class-agnostic mask proposals of common indoor objects from 3D point cloud data using a pre-trained 3D model, while 2D pathway utilizes pre-trained open-vocabulary instance segmentation model to identify a diverse array of object proposals from multi-view RGB-D images. In Dual-Path Integration, our Conditional Integration process, which operates in two stages, filters and merges the proposals from both pathways adaptively. This process harmonizes output proposals to enhance segmentation capabilities. Our framework, utilizing pre-trained models in a zero-shot manner, is model-agnostic and demonstrates superior performance on both seen and unseen data, as evidenced by comprehensive evaluations on the ScanNet200 and qualitative results on ARKitScenes datasets., Comment: OpenSUN 3D: 2nd Workshop on Open-Vocabulary 3D Scene Understanding (CVPR 2024)

This research compares large language model (LLM) fine-tuning methods, including Quantized Low Rank Adapter (QLoRA), Retrieval Augmented fine-tuning (RAFT), and Reinforcement Learning from Human Feedback (RLHF), and additionally compared LLM evaluation methods including End to End (E2E) benchmark method of "Golden Answers", traditional natural language processing (NLP) metrics, RAG Assessment (Ragas), OpenAI GPT-4 evaluation metrics, and human evaluation, using the travel chatbot use case. The travel dataset was sourced from the the Reddit API by requesting posts from travel-related subreddits to get travel-related conversation prompts and personalized travel experiences, and augmented for each fine-tuning method. We used two pretrained LLMs utilized for fine-tuning research: LLaMa 2 7B, and Mistral 7B. QLoRA and RAFT are applied to the two pretrained models. The inferences from these models are extensively evaluated against the aforementioned metrics. The best model according to human evaluation and some GPT-4 metrics was Mistral RAFT, so this underwent a Reinforcement Learning from Human Feedback (RLHF) training pipeline, and ultimately was evaluated as the best model. Our main findings are that: 1) quantitative and Ragas metrics do not align with human evaluation, 2) Open AI GPT-4 evaluation most aligns with human evaluation, 3) it is essential to keep humans in the loop for evaluation because, 4) traditional NLP metrics insufficient, 5) Mistral generally outperformed LLaMa, 6) RAFT outperforms QLoRA, but still needs postprocessing, 7) RLHF improves model performance significantly. Next steps include improving data quality, increasing data quantity, exploring RAG methods, and focusing data collection on a specific city, which would improve data quality by narrowing the focus, while creating a useful product.

Reinforcement Learning (RL) agents demonstrating proficiency in a training environment exhibit vulnerability to adversarial perturbations in input observations during deployment. This underscores the importance of building a robust agent before its real-world deployment. To alleviate the challenging point, prior works focus on developing robust training-based procedures, encompassing efforts to fortify the deep neural network component's robustness or subject the agent to adversarial training against potent attacks. In this work, we propose a novel method referred to as Transformed Input-robust RL (TIRL), which explores another avenue to mitigate the impact of adversaries by employing input transformation-based defenses. Specifically, we introduce two principles for applying transformation-based defenses in learning robust RL agents: (1) autoencoder-styled denoising to reconstruct the original state and (2) bounded transformations (bit-depth reduction and vector quantization (VQ)) to achieve close transformed inputs. The transformations are applied to the state before feeding it into the policy network. Extensive experiments on multiple MuJoCo environments demonstrate that input transformation-based defenses, i.e., VQ, defend against several adversaries in the state observations. The official code is available at https://github.com/tunglm2203/tirl, Comment: 12 pages (Code: https://github.com/tunglm2203/tirl)

Ultracold atoms exquisitely controlled by lasers are the quantum foundation, particularly for sensing, timekeeping, and computing, of state-of-the-art quantum science and technology. However, the laboratory-scale infrastructure for such optical-atomic quantum apparatuses rarely translates into commercial applications. A promising solution is miniaturizing the optical layouts onto a chip-scale device integrated with cold atoms inside a compact ultra-high vacuum (UHV) chamber. For prototyping purposes, however, rapidly loading or exchanging test photonic devices into a UHV chamber is limited by the evacuation time from atmospheric pressures to the optimal pressures for ultracold atoms of $1\times10^{-11}$ Torr, a process typically taking weeks or months without cryogenics. Here, we present a loadlock apparatus and loading procedure capable of venting, exchanging, and evacuating back to $<1\times10^{-11}$ Torr in under 24 hours. Our system allows for rapid testing and benchmarking of various photonic devices with ultracold atoms.

Access to detailed war impact assessments is crucial for humanitarian organizations to effectively assist populations most affected by armed conflicts. However, maintaining a comprehensive understanding of the situation on the ground is challenging, especially in conflicts that cover vast territories and extend over long periods. This study presents a scalable and transferable method for estimating war-induced damage to buildings. We first train a machine learning model to output pixel-wise probability of destruction from Synthetic Aperture Radar (SAR) satellite image time series, leveraging existing, manual damage assessments as ground truth and cloud-based geospatial analysis tools for large-scale inference. We further post-process these assessments using open building footprints to obtain a final damage estimate per building. We introduce an accessible, open-source tool that allows users to adjust the confidence interval based on their specific requirements and use cases. Our approach enables humanitarian organizations and other actors to rapidly screen large geographic regions for war impacts. We provide two publicly accessible dashboards: a Ukraine Damage Explorer to dynamically view our pre-computed estimates, and a Rapid Damage Mapping Tool to easily run our method and produce custom maps.

Crystal plasticity finite element simulations are frequently employed to predict the plastic anisotropy of polycrystalline metals based on their crystallographic texture. In age-hardenable aluminium alloys, however, the texture-induced plastic anisotropy is known to affect by precipitation. This paper presents a new modelling approach to incorporate this effect into crystal plasticity constitutive models. The approach focuses on the overall effect of precipitation, which is assumed to result in an additional directional dependency with respect to a global material orientation, superimposed with the texture-induced plastic anisotropy. This additional directional dependency is implemented into a conventional crystal plasticity constitutive model via a modified hardening law that introduces two new parameters, of which only one is treated as a free parameter. To demonstrate the applicability of the new modelling approach, it is applied to an age-hardenable AA6014-T4 aluminium alloy and compared against a state-of-the-art crystal plasticity constitutive model that considers only crystallographic texture. The results demonstrate that the new modelling approach significantly improves the prediction accuracy of the plastic anisotropy for the AA6014-T4 aluminium alloy studied., Comment: as submitted, after one round of review

Personalized decision making requires the knowledge of potential outcomes under different treatments, and confidence intervals about the potential outcomes further enrich this decision-making process and improve its reliability in high-stakes scenarios. Predicting potential outcomes along with its uncertainty in a counterfactual world poses the foundamental challenge in causal inference. Existing methods that construct confidence intervals for counterfactuals either rely on the assumption of strong ignorability, or need access to un-identifiable lower and upper bounds that characterize the difference between observational and interventional distributions. To overcome these limitations, we first propose a novel approach wTCP-DR based on transductive weighted conformal prediction, which provides confidence intervals for counterfactual outcomes with marginal converage guarantees, even under hidden confounding. With less restrictive assumptions, our approach requires access to a fraction of interventional data (from randomized controlled trials) to account for the covariate shift from observational distributoin to interventional distribution. Theoretical results explicitly demonstrate the conditions under which our algorithm is strictly advantageous to the naive method that only uses interventional data. After ensuring valid intervals on counterfactuals, it is straightforward to construct intervals for individual treatment effects (ITEs). We demonstrate our method across synthetic and real-world data, including recommendation systems, to verify the superiority of our methods compared against state-of-the-art baselines in terms of both coverage and efficiency, Comment: Published in SIGKDD'24

Accurate detection and characterization of nanoparticles within confined spaces is crucial for applications ranging from nanofluidics to biotechnology. We present a novel approach that combines interferometric scattering (iSCAT) detection with trapping by dielectrophoresis (DEP) to achieve label-free detection of nanoparticles that are trapped and/or actuated between nano-gap electrodes. DEP utilizes the interaction between the induced dipole of the particle and the applied electric field to create a trapping potential. We demonstrate our method by trapping and label-free detection of down to 20 nm polystyrene nanoparticles. Additionally, we demonstrate that the signal-to-noise ratio of our detection can be boosted up to 20-fold by periodic actuation of the nanoparticle in the trap. This is done by a digital lock-in detection scheme on the modulated scattering signal. Our method holds promise for various applications, including assembly of nanoparticles, single-particle property analysis, and nanofluidic devices.

Offline reinforcement learning (RL) addresses the challenge of expensive and high-risk data exploration inherent in RL by pre-training policies on vast amounts of offline data, enabling direct deployment or fine-tuning in real-world environments. However, this training paradigm can compromise policy robustness, leading to degraded performance in practical conditions due to observation perturbations or intentional attacks. While adversarial attacks and defenses have been extensively studied in deep learning, their application in offline RL is limited. This paper proposes a framework to enhance the robustness of offline RL models by leveraging advanced adversarial attacks and defenses. The framework attacks the actor and critic components by perturbing observations during training and using adversarial defenses as regularization to enhance the learned policy. Four attacks and two defenses are introduced and evaluated on the D4RL benchmark. The results show the vulnerability of both the actor and critic to attacks and the effectiveness of the defenses in improving policy robustness. This framework holds promise for enhancing the reliability of offline RL models in practical scenarios.

Bismuth vanadate (BiVO4) has emerged as a highly prospective material for photoanodes in photoelectrochemical (PEC) water oxidation. However, current limitations with this material lie in the difficulties in producing stable and continuous BiVO4 layers with efficient carrier transfer kinetics, thereby impeding its widespread application in water splitting processes. This study introduces a new fabrication approach that yields continuous, monoclinic nanostructured BiVO4 films, paving the way for their use as photoanodes in efficient PEC water oxidation for hydrogen production under solar light conditions. The fabrication involves the intercalation of vanadium (V) ions into Bi2O3 films at 450oC. Upon interaction with V ions, the film undergoes a transformation from tetragonal Bi2O3 to monoclinic scheelite BiVO4. This synthesis method enables the fabrication of single monoclinic phase BiVO4 films with thicknesses up to 270 nm. The engineered monoclinic BiVO4 film, devoid of any pinholes that could cause carrier loss, exhibits a robust photocurrent of 1.0 mA/cm2 at 1.23 VRHE in a neutral electrolyte, without requiring additional modifications or doping. Moreover, we demonstrate that the incorporation of a cobalt phosphate (Co-Pi) co-catalyst into the BiVO4 photoanode significantly enhances the lifetime of photogenerated holes by a factor of nine, resulting in a further elevation of the photocurrent to 2.9 mA/cm2. This remarkable PEC enhancement can be attributed to the surface state passivation by the Co-Pi co-catalyst. Our fabrication approach opens up a new facile route for producing large-scale, highly efficient BiVO4 photoanodes for PEC water splitting technology.

The NaI:Tl crystals were early investigated and used for wide application fields due to high light yield and crystal growth advantages. So far, the absolute light yields of NaI:Tl crystal have typically been known to be 40 ph/keV. However, it varies widely, far from the theoretical estimation. Since the high light yield and better sensitivity of NaI:Tl crystal is important for low mass dark matter search. Therefore, it is necessary to use high light NaI:Tl crystal, and absolute light yield should be measured with accuracy. In this work, we use the single photoelectron technique for measuring the absolute light yield of 35 NaI:Tl crystals with various sizes from different vendors. There are several high-quality crystals from the COSINE-100 experiment and commercial companies in these crystals. The theoretical estimation and GEANT4 optical simulation have been studied to investigate the PMT optics. Results show the essential role of this correction in avoiding overrated light yield values. The SPE technique using different PMT was compared to the photodiode and avalanche photodiode methods. A 10% systematic error was obtained. Our results show the excellent absolute light yield of NaI:Tl, at 59.4 +- 5.9 ph/keV, while the theoretical predicted light yield is around 70 ph/keV. The evaluation with NaI:Tl crystals in the COSINE-100 experiment has been performed. The six crystals in the COSINE-100 experiment have a high light yield. Based on our results, the light loss of encapsulation needs to be improved, especially for the big-size crystals., Comment: 14 pages, 16 figures

Accurate weather forecasting is of paramount importance for a wide range of practical applications, drawing substantial scientific and societal interest. However, the intricacies of weather systems pose substantial challenges to accurate predictions. This research introduces a multilayer perceptron model tailored for weather forecasting in Itoshima, Kyushu, Japan. Our meticulously designed architecture demonstrates superior performance compared to existing models, surpassing benchmarks such as Long Short-Term Memory and Recurrent Neural Networks.

We present a methodology for the generative reconstruction of 3D Volume Elements (VE) for numerical multiscale analysis of Ti-6Al-4V processed by Additive Manufacturing (AM). The basketweave morphology, which is typically dominant in AM-processed Ti-6Al-4V, is analyzed in conventional Electron Backscatter Diffusion (EBSD) micrographs. Prior \b{eta}-grain reconstruction is performed to obtain the out-of-plane orientation of the observed grains leveraging Burgers orientation relationship. Convolutional Neural Network (CNN) - based microstructure descriptors are extracted from the 2D data, and used for cross-section-based optimization of pixel values on orthogonal planes in 3D, using the Microstructure Characterization and Reconstruction (MCR) implementation MCRpy [16]. In order to utilize MCRpy, which performs best for binary systems, the basketweave microstructure, which consists of up to twelve distinct grain orientations, is decomposed into several separate two-phase systems. Our reconstructions capture key characteristics of the titanium basketweave morphology and show qualitative resemblance to experimentally obtained 3D data. The preservation of volume fraction during assembly of the reconstruction remains an unadressed challenge at this stage.

We introduce Adversarial Policy Optimization (AdvPO), a novel solution to the pervasive issue of reward over-optimization in Reinforcement Learning from Human Feedback (RLHF) for Large Language Models (LLMs). Over-optimization occurs when a reward model serves as an imperfect proxy for human preference, and RL-driven policy optimization erroneously exploits reward inaccuracies. In this paper, we begin by introducing a lightweight way to quantify uncertainties in rewards, relying solely on the last layer embeddings of the reward model, without the need for computationally expensive reward ensembles. AdvPO then addresses a distributionally robust optimization problem centred around the confidence interval of the reward model's predictions for policy improvement. Through comprehensive experiments on the Anthropic HH and TL;DR summarization datasets, we illustrate the efficacy of AdvPO in mitigating the overoptimization issue, consequently resulting in enhanced performance as evaluated through human-assisted evaluation.

In machine learning fairness, training models that minimize disparity across different sensitive groups often leads to diminished accuracy, a phenomenon known as the fairness-accuracy trade-off. The severity of this trade-off inherently depends on dataset characteristics such as dataset imbalances or biases and therefore, using a uniform fairness requirement across diverse datasets remains questionable. To address this, we present a computationally efficient approach to approximate the fairness-accuracy trade-off curve tailored to individual datasets, backed by rigorous statistical guarantees. By utilizing the You-Only-Train-Once (YOTO) framework, our approach mitigates the computational burden of having to train multiple models when approximating the trade-off curve. Crucially, we introduce a novel methodology for quantifying uncertainty in our estimates, thereby providing practitioners with a robust framework for auditing model fairness while avoiding false conclusions due to estimation errors. Our experiments spanning tabular (e.g., Adult), image (CelebA), and language (Jigsaw) datasets underscore that our approach not only reliably quantifies the optimum achievable trade-offs across various data modalities but also helps detect suboptimality in SOTA fairness methods.

Activists, governmentsm and academics regularly advocate for more open data. But how is data made open, and for whom is it made useful and usable? In this paper, we investigate and describe the work of making eviction data open to tenant organizers. We do this through an ethnographic description of ongoing work with a local housing activist organization. This work combines observation, direct participation in data work, and creating media artifacts, specifically digital maps. Our interpretation is grounded in D'Ignazio and Klein's Data Feminism, emphasizing standpoint theory. Through our analysis and discussion, we highlight how shifting positionalities from data intermediaries to data accomplices affects the design of data sets and maps. We provide HCI scholars with three design implications when situating data for grassroots organizers: becoming a domain beginner, striving for data actionability, and evaluating our design artifacts by the social relations they sustain rather than just their technical efficacy., Comment: 16 pages including references, 4 figures, 1 table, ACM CHI 2024

Cheating in online exams has become a prevalent issue over the past decade, especially during the COVID-19 pandemic. To address this issue of academic dishonesty, our "Exam Monitoring System: Detecting Abnormal Behavior in Online Examinations" is designed to assist proctors in identifying unusual student behavior. Our system demonstrates high accuracy and speed in detecting cheating in real-time scenarios, providing valuable information, and aiding proctors in decision-making. This article outlines our methodology and the effectiveness of our system in mitigating the widespread problem of cheating in online exams.

LLM hallucination, i.e. generating factually incorrect yet seemingly convincing answers, is currently a major threat to the trustworthiness and reliability of LLMs. The first step towards solving this complicated problem is to measure it. However, existing hallucination metrics require having a benchmark dataset with gold-standard answers, i.e. "best" or "correct" answers written by humans. Such requirements make hallucination measurement costly and prone to human errors. In this work, we propose Factualness Evaluations via Weighting LLMs (FEWL), an innovative hallucination metric that is specifically designed for the scenario when gold-standard answers are absent. FEWL leverages the answers from off-the-shelf LLMs that serve as a proxy of gold-standard answers. The key challenge is how to quantify the expertise of reference LLMs resourcefully. We show FEWL has certain theoretical guarantees and demonstrate empirically it gives more accurate hallucination measures than naively using reference LLMs. We also show how to leverage FEWL to reduce hallucination through both in-context learning and supervised fine-tuning. Extensive experiment results on Truthful-QA, CHALE, and HaluEval datasets demonstrate the effectiveness of FEWL., Comment: Paper Under Review

Adding superconducting (SC) electron pairing via the proximity effect to pristinely non-superconducting materials can lead to a variety of interesting physical phenomena. Particular interest has recently focused on inducing SC into two-dimensional surface states (SSs), potentially also combined with non-trivial topology. We study the mechanism of proximity-induced SC into the Shockley-type SSs of the noble metals Ag(111) and Cu(111) grown on the elemental SC Nb(110) using scanning tunneling spectroscopy. The tunneling spectra exhibit an intriguing multitude of sharp states at low energies. Their appearance can be explained by Andreev bound states (ABS) formed by the weakly proximitized SSs subject to lateral finite-size confinement. We study systematically how the proximity gap in the bulk states of both Ag(111) and Cu(111) persists up to island thicknesses of several times the bulk coherence length of Nb. We find that even for thick islands, the SSs acquire a gap, with the gap size for Cu being consistently larger than for Ag. Based on this, we argue that the SC in the SS is not provided through direct overlap of the SS wavefunction with the SC host but can be understood to be mediated by step edges inducing electronic coupling to the bulk. Our work provides important input for the microscopic understanding of induced superconductivity in heterostructures and its spectral manifestation. Moreover, it lays the foundation for more complex SC heterostructures based on noble metals.

LGBTQ+ people have received increased attention in HCI research, paralleling a greater emphasis on social justice in recent years. However, there has not been a systematic review of how LGBTQ+ people are researched or discussed in HCI. In this work, we review all research mentioning LGBTQ+ people across the HCI venues of CHI, CSCW, DIS, and TOCHI. Since 2014, we find a linear growth in the number of papers substantially about LGBTQ+ people and an exponential increase in the number of mentions. Research about LGBTQ+ people tends to center experiences of being politicized, outside the norm, stigmatized, or highly vulnerable. LGBTQ+ people are typically mentioned as a marginalized group or an area of future research. We identify gaps and opportunities for (1) research about and (2) the discussion of LGBTQ+ in HCI and provide a dataset to facilitate future Queer HCI research.