Search

Multimodal foundation models, such as Gemini and ChatGPT, have revolutionized human-machine interactions by seamlessly integrating various forms of data. Developing a universal spoken language model that comprehends a wide range of natural language instructions is critical for bridging communication gaps and facilitating more intuitive interactions. However, the absence of a comprehensive evaluation benchmark poses a significant challenge. We present Dynamic-SUPERB Phase-2, an open and evolving benchmark for the comprehensive evaluation of instruction-based universal speech models. Building upon the first generation, this second version incorporates 125 new tasks contributed collaboratively by the global research community, expanding the benchmark to a total of 180 tasks, making it the largest benchmark for speech and audio evaluation. While the first generation of Dynamic-SUPERB was limited to classification tasks, Dynamic-SUPERB Phase-2 broadens its evaluation capabilities by introducing a wide array of novel and diverse tasks, including regression and sequence generation, across speech, music, and environmental audio. Evaluation results indicate that none of the models performed well universally. SALMONN-13B excelled in English ASR, while WavLLM demonstrated high accuracy in emotion recognition, but current models still require further innovations to handle a broader range of tasks. We will soon open-source all task data and the evaluation pipeline.

Jointly optimizing power allocation and device association is crucial in Internet-of-Things (IoT) networks to ensure devices achieve their data throughput requirements. Device association, which assigns IoT devices to specific access points (APs), critically impacts resource allocation. Many existing works often assume all data throughput requirements are satisfied, which is impractical given resource limitations and diverse demands. When requirements cannot be met, the system becomes infeasible, causing congestion and degraded performance. To address this problem, we propose a novel framework to enhance IoT system robustness by solving two problems, comprising maximizing the number of satisfied IoT devices and jointly maximizing both the number of satisfied devices and total network throughput. These objectives often conflict under infeasible circumstances, necessitating a careful balance. We thus propose a modified branch-and-bound (BB)-based method to solve the first problem. An iterative algorithm is proposed for the second problem that gradually increases the number of satisfied IoT devices and improves the total network throughput. We employ a logarithmic approximation for a lower bound on data throughput and design a fixed-point algorithm for power allocation, followed by a coalition game-based method for device association. Numerical results demonstrate the efficiency of the proposed algorithm, serving fewer devices than the BB-based method but with faster running time and higher total throughput., Comment: 18 pages, 8 figures, and 4 tables. Accepted by IEEE Transactions on Network and Service Management

Audience feedback is crucial for refining video content, yet it typically comes after publication, limiting creators' ability to make timely adjustments. To bridge this gap, we introduce SimTube, a generative AI system designed to simulate audience feedback in the form of video comments before a video's release. SimTube features a computational pipeline that integrates multimodal data from the video-such as visuals, audio, and metadata-with user personas derived from a broad and diverse corpus of audience demographics, generating varied and contextually relevant feedback. Furthermore, the system's UI allows creators to explore and customize the simulated comments. Through a comprehensive evaluation-comprising quantitative analysis, crowd-sourced assessments, and qualitative user studies-we show that SimTube's generated comments are not only relevant, believable, and diverse but often more detailed and informative than actual audience comments, highlighting its potential to help creators refine their content before release.

Retrieval-augmented generation (RAG) has emerged as a promising approach to enhance the performance of large language models (LLMs) in knowledge-intensive tasks such as those from medical domain. However, the sensitive nature of the medical domain necessitates a completely accurate and trustworthy system. While existing RAG benchmarks primarily focus on the standard retrieve-answer setting, they overlook many practical scenarios that measure crucial aspects of a reliable medical system. This paper addresses this gap by providing a comprehensive evaluation framework for medical question-answering (QA) systems in a RAG setting for these situations, including sufficiency, integration, and robustness. We introduce Medical Retrieval-Augmented Generation Benchmark (MedRGB) that provides various supplementary elements to four medical QA datasets for testing LLMs' ability to handle these specific scenarios. Utilizing MedRGB, we conduct extensive evaluations of both state-of-the-art commercial LLMs and open-source models across multiple retrieval conditions. Our experimental results reveals current models' limited ability to handle noise and misinformation in the retrieved documents. We further analyze the LLMs' reasoning processes to provides valuable insights and future directions for developing RAG systems in this critical medical domain.

Vision Transformers (ViTs) have shown exceptional performance in vehicle re-identification (ReID) tasks. However, non-square aspect ratios of image or video inputs can negatively impact re-identification accuracy. To address this challenge, we propose a novel, human perception driven, and general ViT-based ReID framework that fuses models trained on various aspect ratios. Our key contributions are threefold: (i) We analyze the impact of aspect ratios on performance using the VeRi-776 and VehicleID datasets, providing guidance for input settings based on the distribution of original image aspect ratios. (ii) We introduce patch-wise mixup strategy during ViT patchification (guided by spatial attention scores) and implement uneven stride for better alignment with object aspect ratios. (iii) We propose a dynamic feature fusion ReID network to enhance model robustness. Our method outperforms state-of-the-art transformer-based approaches on both datasets, with only a minimal increase in inference time per image.

Ram pressure stripping (RPS) is an important process that plays a significant role in shaping the evolution of cluster galaxies and their surrounding environment. Despite its recognized significance, the potential connection between RPS and AGN activity in cluster galaxies remains poorly understood. Recent claims, based on optical emission line diagnostics, have suggested such a connection. Here, we investigate this relationship from an X-ray perspective using a sample of galaxies undergoing RPS in four nearby galaxy clusters - A1656, A1367, A426, and A3627. This study is the first to test such a connection from an X-ray standpoint. Our analysis reveals no signs of enhanced X-ray AGN activity in our sample, with most RPS galaxies ($\sim$ $90\%$) showing X-ray luminosities below $10^{41}$ erg s$^{-1}$ in their central point sources. Moreover, there is no noticeable difference in X-ray AGN activity among RPS galaxies compared to a control sample of non-RPS galaxies, as demonstrated by similar X-ray luminosities observed in their central point sources. While the most luminous X-ray AGN in our sample is found in ESO 137-002, a galaxy undergoing RPS in A3627, there is no evidence for a widespread enhancement of X-ray AGN activity due to RPS. Given the limited sample size of our study, this could also indicate that either the X-ray AGN enhancement from RPS is at most weak, or the timescale for the X-ray AGN enhancement is short. This emphasizes the need for further investigations with larger X-ray samples to better understand the impact of RPS on AGN activity in cluster galaxies., Comment: 17 pages, 3 figures, 2 tables, revised and resubmitted to The Astrophysical Journal

Many nations are promoting the green transition in the energy sector to attain neutral carbon emissions by 2050. Smart Grid 2.0 (SG2) is expected to explore data-driven analytics and enhance communication technologies to improve the efficiency and sustainability of distributed renewable energy systems. These features are beyond smart metering and electric surplus distribution in conventional smart grids. Given the high dependence on communication networks to connect distributed microgrids in SG2, potential cascading failures of connectivity can cause disruption to data synchronization to the remote control systems. This paper reviews security threats and defense tactics for three stakeholders: power grid operators, communication network providers, and consumers. Through the survey, we found that SG2's stakeholders are particularly vulnerable to substation attacks/vandalism, malware/ransomware threats, blockchain vulnerabilities and supply chain breakdowns. Furthermore, incorporating artificial intelligence (AI) into autonomous energy management in distributed energy resources of SG2 creates new challenges. Accordingly, adversarial samples and false data injection on electricity reading and measurement sensors at power plants can fool AI-powered control functions and cause messy error-checking operations in energy storage, wrong energy estimation in electric vehicle charging, and even fraudulent transactions in peer-to-peer energy trading models. Scalable blockchain-based models, physical unclonable function, interoperable security protocols, and trustworthy AI models designed for managing distributed microgrids in SG2 are typical promising protection models for future research., Comment: 30 pages, 21 figures, 5 tables, accepted to appear in IEEE COMST

This paper proposes a distributed learning-based framework to tackle the sum ergodic rate maximization problem in cell-free massive multiple-input multiple-output (MIMO) systems by utilizing the graph neural network (GNN). Different from centralized schemes, which gather all the channel state information (CSI) at the central processing unit (CPU) for calculating the resource allocation, the local resource of access points (APs) is exploited in the proposed distributed GNN-based framework to allocate transmit powers. Specifically, APs can use a unique GNN model to allocate their power based on the local CSI. The GNN model is trained at the CPU using the local CSI of one AP, with partially exchanged information from other APs to calculate the loss function to reflect system characteristics, capturing comprehensive network information while avoiding computation burden. Numerical results show that the proposed distributed learning-based approach achieves a sum ergodic rate close to that of centralized learning while outperforming the model-based optimization., Comment: 6 pages, 4 figures, and 4 tables. Accepted by IEEE TVT

Directed acyclic graphs (DAGs) serve as crucial data representations in domains such as hardware synthesis and compiler/program optimization for computing systems. DAG generative models facilitate the creation of synthetic DAGs, which can be used for benchmarking computing systems while preserving intellectual property. However, generating realistic DAGs is challenging due to their inherent directional and logical dependencies. This paper introduces LayerDAG, an autoregressive diffusion model, to address these challenges. LayerDAG decouples the strong node dependencies into manageable units that can be processed sequentially. By interpreting the partial order of nodes as a sequence of bipartite graphs, LayerDAG leverages autoregressive generation to model directional dependencies and employs diffusion models to capture logical dependencies within each bipartite graph. Comparative analyses demonstrate that LayerDAG outperforms existing DAG generative models in both expressiveness and generalization, particularly for generating large-scale DAGs with up to 400 nodes-a critical scenario for system benchmarking. Extensive experiments on both synthetic and real-world flow graphs from various computing platforms show that LayerDAG generates valid DAGs with superior statistical properties and benchmarking performance. The synthetic DAGs generated by LayerDAG enhance the training of ML-based surrogate models, resulting in improved accuracy in predicting performance metrics of real-world DAGs across diverse computing platforms., Comment: Code available at https://github.com/Graph-COM/LayerDAG

Customer Relationship Management (CRM) systems are vital for modern enterprises, providing a foundation for managing customer interactions and data. Integrating AI agents into CRM systems can automate routine processes and enhance personalized service. However, deploying and evaluating these agents is challenging due to the lack of realistic benchmarks that reflect the complexity of real-world CRM tasks. To address this issue, we introduce CRMArena, a novel benchmark designed to evaluate AI agents on realistic tasks grounded in professional work environments. Following guidance from CRM experts and industry best practices, we designed CRMArena with nine customer service tasks distributed across three personas: service agent, analyst, and manager. The benchmark includes 16 commonly used industrial objects (e.g., account, order, knowledge article, case) with high interconnectivity, along with latent variables (e.g., complaint habits, policy violations) to simulate realistic data distributions. Experimental results reveal that state-of-the-art LLM agents succeed in less than 40% of the tasks with ReAct prompting, and less than 55% even with function-calling abilities. Our findings highlight the need for enhanced agent capabilities in function-calling and rule-following to be deployed in real-world work environments. CRMArena is an open challenge to the community: systems that can reliably complete tasks showcase direct business value in a popular work environment.

The ability of intelligent systems to predict human behaviors is crucial, particularly in fields such as autonomous vehicle navigation and social robotics. However, the complexity of human motion have prevented the development of a standardized dataset for human motion prediction, thereby hindering the establishment of pre-trained models. In this paper, we address these limitations by integrating multiple datasets, encompassing both trajectory and 3D pose keypoints, to propose a pre-trained model for human motion prediction. We merge seven distinct datasets across varying modalities and standardize their formats. To facilitate multimodal pre-training, we introduce Multi-Transmotion, an innovative transformer-based model designed for cross-modality pre-training. Additionally, we present a novel masking strategy to capture rich representations. Our methodology demonstrates competitive performance across various datasets on several downstream tasks, including trajectory prediction in the NBA and JTA datasets, as well as pose prediction in the AMASS and 3DPW datasets. The code is publicly available: https://github.com/vita-epfl/multi-transmotion, Comment: CoRL 2024

Student mental health is a sensitive issue that necessitates special attention. A primary concern is the student-to-counselor ratio, which surpasses the recommended standard of 250:1 in most universities. This imbalance results in extended waiting periods for in-person consultations, which cause suboptimal treatment. Significant efforts have been directed toward developing mental health dialogue systems utilizing the existing open-source mental health-related datasets. However, currently available datasets either discuss general topics or various strategies that may not be viable for direct application due to numerous ethical constraints inherent in this research domain. To address this issue, this paper introduces a specialized mental health dataset that emphasizes the active listening strategy employed in conversation for counseling, also named as ConvCounsel. This dataset comprises both speech and text data, which can facilitate the development of a reliable pipeline for mental health dialogue systems. To demonstrate the utility of the proposed dataset, this paper also presents the NYCUKA, a spoken mental health dialogue system that is designed by using the ConvCounsel dataset. The results show the merit of using this dataset., Comment: Accepted at O-COCOSDA 2024, Won Best Student Paper Award

Parkinson's disease (PD), a degenerative disorder of the central nervous system, is commonly diagnosed using functional medical imaging techniques such as single-photon emission computed tomography (SPECT). In this study, we utilized two SPECT data sets (n = 634 and n = 202) from different hospitals to develop a model capable of accurately predicting PD stages, a multiclass classification task. We used the entire three-dimensional (3D) brain images as input and experimented with various model architectures. Initially, we treated the 3D images as sequences of two-dimensional (2D) slices and fed them sequentially into 2D convolutional neural network (CNN) models pretrained on ImageNet, averaging the outputs to obtain the final predicted stage. We also applied 3D CNN models pretrained on Kinetics-400. Additionally, we incorporated an attention mechanism to account for the varying importance of different slices in the prediction process. To further enhance model efficacy and robustness, we simultaneously trained the two data sets using weight sharing, a technique known as cotraining. Our results demonstrated that 2D models pretrained on ImageNet outperformed 3D models pretrained on Kinetics-400, and models utilizing the attention mechanism outperformed both 2D and 3D models. The cotraining technique proved effective in improving model performance when the cotraining data sets were sufficiently large., Comment: 34 pages, 7 figures, and 4 tables

As large language models (LLMs) expand into performing as agents for real-world applications beyond traditional NLP tasks, evaluating their robustness becomes increasingly important. However, existing benchmarks often overlook critical dimensions like cultural and social awareness. To address these, we introduce CASA, a benchmark designed to assess LLM agents' sensitivity to cultural and social norms across two web-based tasks: online shopping and social discussion forums. Our approach evaluates LLM agents' ability to detect and appropriately respond to norm-violating user queries and observations. Furthermore, we propose a comprehensive evaluation framework that measures awareness coverage, helpfulness in managing user queries, and the violation rate when facing misleading web content. Experiments show that current LLMs perform significantly better in non-agent than in web-based agent environments, with agents achieving less than 10% awareness coverage and over 40% violation rates. To improve performance, we explore two methods: prompting and fine-tuning, and find that combining both methods can offer complementary advantages -- fine-tuning on culture-specific datasets significantly enhances the agents' ability to generalize across different regions, while prompting boosts the agents' ability to navigate complex tasks. These findings highlight the importance of constantly benchmarking LLM agents' cultural and social awareness during the development cycle., Comment: Work in progress

With physical quantum computers becoming increasingly accessible, research on their applications across various fields has advanced rapidly. In this paper, we present the first study of quantum cosmology conducted on physical quantum computers, employing a newly proposed Hybrid Quantum-Classical (HQC) algorithm rather than the commonly used Variational Quantum Eigensolver (VQE). Specifically, we solve a constrained Hamiltonian equation derived by quantizing the Friedmann equation in cosmology. To solve this constraint equation, H |psi> = 0, where H is a Hamiltonian operator and |psi> = |psi(theta)> is the wave function of phase angle theta describing the cosmic quantum state, we iteratively use the quantum computer to compute the eigenvalues of , while a classical computer manages the underlying probability density function within the Probabilistic Bisection Algorithm (PBA) to update theta until the solution of = 0 is achieved to a desired accuracy. Executing our algorithm on IBM's quantum computers, we attain a high-precision solution for theta, achieving approximately 1 percent error., Comment: 8 pages, 4 figure

Chirality is ubiquitous in nature and manifests in a wide range of phenomena including chemical reactions, biological processes, and quantum transport of electrons. In quantum materials, the chirality of fermions, given by the relative directions between the electron spin and momentum, is connected to the band topology of electronic states. Here, we show that in structurally chiral materials like CoSi, the orbital angular momentum (OAM) serves as the main driver of a nontrivial band topology in this new class of unconventional topological semimetals, even when spin-orbit coupling is negligible. A nontrivial orbital-momentum locking of multifold chiral fermions in the bulk leads to a pronounced OAM texture of the helicoid Fermi arcs at the surface. Our findings highlight the pivotal role of the orbital degree of freedom for the chirality and topology of electron states, in general, and pave the way towards the application of topological chiral semimetals in orbitronic devices.

Synchrotron X-ray emission has been detected from nearly a dozen young supernova remnants (SNRs). X-rays of synchrotron origin exhibit linear polarization in a regular, non-randomly oriented magnetic field. The significant polarized X-ray emission from four such SNRs has already been reported on the basis of observations with the Imaging X-ray Polarimetry Explorer (IXPE). The magnetic-field structure as derived from IXPE observations is radial for Cassiopeia A, Tycho's SNR, and SN 1006, and tangential for RX J1713.7-3946. The latter together with the recent detection of a tangential magnetic field in SNR 1E 0102.2-7219 by the Australia Telescope Compact Array in the radio band shows that tangential magnetic fields can also be present in young SNRs. Thus, the dichotomy in polarization between young and middle-aged SNRs (radial magnetic fields in young SNRs, but tangential magnetic fields in middle-aged SNRs), previously noticed in the radio band, deserves additional attention. The present analysis of IXPE observations determines, for the first time, a magnetic-field structure in the northwestern rim of Vela Jr, also known as RX J0852.0-4622, and provides a new example of a young SNR with a tangential magnetic field., Comment: Accepted for publication in Astronomy and Astrophysics

Autoencoder permits the end-to-end optimization and design of wireless communication systems to be more beneficial than traditional signal processing. However, this emerging learning-based framework has weaknesses, especially sensitivity to physical attacks. This paper explores adversarial attacks against a double reconfigurable intelligent surface (RIS)-assisted multiple-input and multiple-output (MIMO)-based autoencoder, where an adversary employs encoded and decoded datasets to create adversarial perturbation and fool the system. Because of the complex and dynamic data structures, adversarial attacks are not unique, each having its own benefits. We, therefore, propose three algorithms generating adversarial examples and perturbations to attack the RIS-MIMO-based autoencoder, exploiting the gradient descent and allowing for flexibility via varying the input dimensions. Numerical results show that the proposed adversarial attack-based algorithm significantly degrades the system performance regarding the symbol error rate compared to the jamming attacks., Comment: 5 pages, 2 figures. Accepted by WCL

Small Language Models (SLMs) have become increasingly important due to their efficiency and performance to perform various language tasks with minimal computational resources, making them ideal for various settings including on-device, mobile, edge devices, among many others. In this article, we present a comprehensive survey on SLMs, focusing on their architectures, training techniques, and model compression techniques. We propose a novel taxonomy for categorizing the methods used to optimize SLMs, including model compression, pruning, and quantization techniques. We summarize the benchmark datasets that are useful for benchmarking SLMs along with the evaluation metrics commonly used. Additionally, we highlight key open challenges that remain to be addressed. Our survey aims to serve as a valuable resource for researchers and practitioners interested in developing and deploying small yet efficient language models.

X-ray polarization is a unique new probe of the particle acceleration in astrophysical jets made possible through the Imaging X-ray Polarimetry Explorer. Here we report on the first dense X-ray polarization monitoring campaign on the blazar Mrk 421. Our observations were accompanied by an even denser radio and optical polarization campaign. We find significant short-timescale variability in both X-ray polarization degree and angle, including a $\sim90^\circ$ angle rotation about the jet axis. We attribute this to random variations of the magnetic field, consistent with the presence of turbulence but also unlikely to be explained by turbulence alone. At the same time, the degree of lower-energy polarization is significantly lower and shows no more than mild variability. Our campaign provides further evidence for a scenario in which energy-stratified shock-acceleration of relativistic electrons, combined with a turbulent magnetic field, is responsible for optical to X-ray synchrotron emission in blazar jets., Comment: 23 pages, including 8 pages of appendices. 12 figures, 3 tables. Submitted to ApJ

Efficient long-context language modeling remains a significant challenge in Natural Language Processing (NLP). While Transformers dominate language tasks, they struggle with long sequences due to quadratic computational complexity in training and linearly scaling memory costs during inference. Recent State Space Models (SSMs) such as Mamba offer alternatives with constant memory usage, but they underperform in tasks requiring extensive in-context retrieval. We introduce Taipan, a novel hybrid architecture that combines Mamba-2 with Selective Attention Layers (SALs). These SALs identify tokens requiring long-range interactions, remove less important features, and then augment their representations using the attention module. This approach balances Mamba's efficiency with Transformer-like performance in memory-intensive tasks. By constraining the attention budget, Taipan extends accurate predictions to context lengths of up to 1 million tokens while preserving computational efficiency. Our experiments demonstrate Taipan's superior performance across various scales and tasks, offering a promising solution for efficient long-context language modeling.

Effective control of magnetic phases in two-dimensional magnets would constitute crucial progress in spintronics, holding great potential for future computing technologies. Here, we report a new approach of leveraging tunneling current as a tool for controlling spin states in CrI3. We reveal that a tunneling current can deterministically switch between spin-parallel and spin-antiparallel states in few-layer CrI3, depending on the polarity and amplitude of the current. We propose a mechanism involving nonequilibrium spin accumulation in the graphene electrodes in contact with the CrI3 layers. We further demonstrate tunneling current-tunable stochastic switching between multiple spin states of the CrI3 tunnel devices, which goes beyond conventional bi-stable stochastic magnetic tunnel junctions and has not been documented in two-dimensional magnets. Our findings not only address the existing knowledge gap concerning the influence of tunneling currents in controlling the magnetism in two-dimensional magnets, but also unlock possibilities for energy-efficient probabilistic and neuromorphic computing., Comment: 23 pages, 4 figures

The classification of medical images is a pivotal aspect of disease diagnosis, often enhanced by deep learning techniques. However, traditional approaches typically focus on unimodal medical image data, neglecting the integration of diverse non-image patient data. This paper proposes a novel Cross-Graph Modal Contrastive Learning (CGMCL) framework for multimodal medical image classification. The model effectively integrates both image and non-image data by constructing cross-modality graphs and leveraging contrastive learning to align multimodal features in a shared latent space. An inter-modality feature scaling module further optimizes the representation learning process by reducing the gap between heterogeneous modalities. The proposed approach is evaluated on two datasets: a Parkinson's disease (PD) dataset and a public melanoma dataset. Results demonstrate that CGMCL outperforms conventional unimodal methods in accuracy, interpretability, and early disease prediction. Additionally, the method shows superior performance in multi-class melanoma classification. The CGMCL framework provides valuable insights into medical image classification while offering improved disease interpretability and predictive capabilities.

As the signatories of this manifesto, we denounce the attention economy as inhumane and a threat to our sociopolitical and ecological well-being. We endorse policymakers' efforts to address the negative consequences of the attention economy's technology, but add that these approaches are often limited in their criticism of the systemic context of human attention. Starting from Buddhist philosophy, we advocate a broader approach: an ecology of attending, that centers on conceptualizing, designing, and using attention (1) in an embedded way and (2) focused on the alleviating of suffering. With 'embedded' we mean that attention is not a neutral, isolated mechanism but a meaning-engendering part of an 'ecology' of bodily, sociotechnical and moral frameworks. With 'focused on the alleviation of suffering' we explicitly move away from the (often implicit) conception of attention as a tool for gratifying desires., Comment: 21 pages, 1 figure

Knowledge graphs (KGs) generated by large language models (LLMs) are becoming increasingly valuable for Retrieval-Augmented Generation (RAG) applications that require knowledge-intensive reasoning. However, existing KG extraction methods predominantly rely on prompt-based approaches, which are inefficient for processing large-scale corpora. These approaches often suffer from information loss, particularly with long documents, due to the lack of specialized design for KG construction. Additionally, there is a gap in evaluation datasets and methodologies for ontology-free KG construction. To overcome these limitations, we propose SynthKG, a multi-step, document-level ontology-free KG synthesis workflow based on LLMs. By fine-tuning a smaller LLM on the synthesized document-KG pairs, we streamline the multi-step process into a single-step KG generation approach called Distill-SynthKG, substantially reducing the number of LLM inference calls. Furthermore, we re-purpose existing question-answering datasets to establish KG evaluation datasets and introduce new evaluation metrics. Using KGs produced by Distill-SynthKG, we also design a novel graph-based retrieval framework for RAG. Experimental results demonstrate that Distill-SynthKG not only surpasses all baseline models in KG quality -- including models up to eight times larger -- but also consistently excels in retrieval and question-answering tasks. Our proposed graph retrieval framework also outperforms all KG-retrieval methods across multiple benchmark datasets. We release the SynthKG dataset and Distill-SynthKG model publicly to support further research and development.

Evaluating retrieval-augmented generation (RAG) systems remains challenging, particularly for open-ended questions that lack definitive answers and require coverage of multiple sub-topics. In this paper, we introduce a novel evaluation framework based on sub-question coverage, which measures how well a RAG system addresses different facets of a question. We propose decomposing questions into sub-questions and classifying them into three types -- core, background, and follow-up -- to reflect their roles and importance. Using this categorization, we introduce a fine-grained evaluation protocol that provides insights into the retrieval and generation characteristics of RAG systems, including three commercial generative answer engines: You.com, Perplexity AI, and Bing Chat. Interestingly, we find that while all answer engines cover core sub-questions more often than background or follow-up ones, they still miss around 50% of core sub-questions, revealing clear opportunities for improvement. Further, sub-question coverage metrics prove effective for ranking responses, achieving 82% accuracy compared to human preference annotations. Lastly, we also demonstrate that leveraging core sub-questions enhances both retrieval and answer generation in a RAG system, resulting in a 74% win rate over the baseline that lacks sub-questions.

Identifying and localizing objects within images is a fundamental challenge, and numerous efforts have been made to enhance model accuracy by experimenting with diverse architectures and refining training strategies. Nevertheless, a prevalent limitation in existing models is overemphasizing the current input while ignoring the information from the entire dataset. We introduce an innovative {\em \textbf{R}etriever}-{\em\textbf{D}ictionary} (RD) module to address this issue. This architecture enables YOLO-based models to efficiently retrieve features from a Dictionary that contains the insight of the dataset, which is built by the knowledge from Visual Models (VM), Large Language Models (LLM), or Visual Language Models (VLM). The flexible RD enables the model to incorporate such explicit knowledge that enhances the ability to benefit multiple tasks, specifically, segmentation, detection, and classification, from pixel to image level. The experiments show that using the RD significantly improves model performance, achieving more than a 3\% increase in mean Average Precision for object detection with less than a 1\% increase in model parameters. Beyond 1-stage object detection models, the RD module improves the effectiveness of 2-stage models and DETR-based architectures, such as Faster R-CNN and Deformable DETR

Two-dimensional (2D) kagome lattice metals are interesting because they display flat electronic bands, Dirac points, Van Hove singularities, and can have interplay between charge density wave (CDW), magnetic order, and superconductivity. In kagome lattice antiferromagnet FeGe, a short-range CDW order was found deep within an antiferromagnetically ordered state, interacting with the magnetic order. Surprisingly, post-growth annealing of FeGe at 560$^{\circ}$C can suppress the CDW order while annealing at 320$^{\circ}$C induces a long-range CDW order, with the ability to cycle between the states repeatedly by annealing. Here we perform transport, neutron scattering, scanning transmission electron microscopy (STEM), and muon spin rotation ($\mu$SR) experiments to unveil the microscopic mechanism of the annealing process and its impact on magneto-transport, CDW, and magnetic properties of FeGe. We find that 560$^{\circ}$C annealing creates germanium vacancies uniformly distributed throughout the FeGe kagome lattice, which prevent the formation of Ge-Ge dimers necessary for the CDW order. Upon annealing at 320$^{\circ}$C, the system segregates into stoichiometric FeGe regions with long-range CDW order and regions with stacking faults that act as nucleation sites for the CDW. The presence or absence of CDW order greatly affects the anomalous Hall effect, incommensurate magnetic order, and spin-lattice coupling in FeGe, thus placing FeGe as the only known kagome lattice material with a tunable CDW and magnetic order, potentially useful for sensing and information transmission.

Large Language Model (LLM)-based applications are graduating from research prototypes to products serving millions of users, influencing how people write and consume information. A prominent example is the appearance of Answer Engines: LLM-based generative search engines supplanting traditional search engines. Answer engines not only retrieve relevant sources to a user query but synthesize answer summaries that cite the sources. To understand these systems' limitations, we first conducted a study with 21 participants, evaluating interactions with answer vs. traditional search engines and identifying 16 answer engine limitations. From these insights, we propose 16 answer engine design recommendations, linked to 8 metrics. An automated evaluation implementing our metrics on three popular engines (You.com, Perplexity.ai, BingChat) quantifies common limitations (e.g., frequent hallucination, inaccurate citation) and unique features (e.g., variation in answer confidence), with results mirroring user study insights. We release our Answer Engine Evaluation benchmark (AEE) to facilitate transparent evaluation of LLM-based applications.

Energy-efficient software helps improve mobile device experiences and reduce the carbon footprint of data centers. However, energy goals are often de-prioritized in order to meet other requirements. We take inspiration from recent work exploring the use of large language models (LLMs) for different software engineering activities. We propose a novel application of LLMs: as code optimizers for energy efficiency. We describe and evaluate a prototype, finding that over 6 small programs our system can improve energy efficiency in 3 of them, up to 2x better than compiler optimizations alone. From our experience, we identify some of the challenges of energy-efficient LLM code optimization and propose a research agenda.

In the Taiwanese judicial system, Cause of Actions (COAs) are essential for identifying relevant legal judgments. However, the lack of standardized COA labeling creates challenges in filtering cases using basic methods. This research addresses this issue by leveraging embedding and clustering techniques to analyze the similarity between COAs based on cited legal articles. The study implements various similarity measures, including Dice coefficient and Pearson's correlation coefficient. An ensemble model combines rankings, and social network analysis identifies clusters of related COAs. This approach enhances legal analysis by revealing inconspicuous connections between COAs, offering potential applications in legal research beyond civil law., Comment: 10 pages, 4 figures, 3 tables(including appendix)

Graphs offer unique insights into relationships and interactions between entities, complementing data modalities like text, images, and videos. By incorporating relational information from graph data, AI models can extend their capabilities beyond traditional tasks. However, relational data in sensitive domains such as finance and healthcare often contain private information, making privacy preservation crucial. Existing privacy-preserving methods, such as DP-SGD, which rely on gradient decoupling assumptions, are not well-suited for relational learning due to the inherent dependencies between coupled training samples. To address this challenge, we propose a privacy-preserving relational learning pipeline that decouples dependencies in sampled relations during training, ensuring differential privacy through a tailored application of DP-SGD. We apply this method to fine-tune large language models (LLMs) on sensitive graph data, and tackle the associated computational complexities. Our approach is evaluated on LLMs of varying sizes (e.g., BERT, Llama2) using real-world relational data from four text-attributed graphs. The results demonstrate significant improvements in relational learning tasks, all while maintaining robust privacy guarantees during training. Additionally, we explore the trade-offs between privacy, utility, and computational efficiency, offering insights into the practical deployment of our approach. Code is available at https://github.com/Graph-COM/PvGaLM.

This paper is a preliminary study on the security and design of cryptosystems using Gabidulin-Kronecker Product Codes. In particular, we point out the design impracticality of the system, and propose ways to improve it.

Recent models suggest approximately half of all accreting supermassive black holes (SMBHs; $M_{\rm BH}$ $\gtrsim$ 10$^{5}$ M$_{\odot}$) are expected to undergo intense growth phases behind Compton-thick ($N_{\rm H}$ $>$ 1.5 $\times$ 10$^{24}$ cm$^{-2}$) veils of obscuring gas. However, despite being a viable source for the seeding of SMBHs, there are currently no examples known of a Compton-thick accreting intermediate mass black hole (IMBH; $M_{\rm BH}$ $\sim$ 10$^{2}$ $-$ 10$^{5}$ M$_{\odot}$). We present a detailed X-ray spectral analysis of IC 750 $-$ the only AGN to-date with a precise megamaser-based intermediate mass $<$ 10$^{5}$ M$_{\odot}$. We find the equivalent width of neutral 6.4 keV Fe K$\alpha$ to be 1.9$^{+2.2}_{-1.0}$ keV via phenomenological modelling of the co-added 177 ks Chandra spectrum. Such large equivalent widths are seldom produced by processes other than fluorescence from dense obscuration. We fit three physically-motivated X-ray spectral models to infer a range of possible intrinsic 2$-$10 keV luminosity posteriors that encompass the systematic uncertainties associated with a choice of model. Despite a wide range of predicted intrinsic 2$-$10 keV luminosities between $\sim$ 10$^{41}$ and 10$^{43}$ erg s$^{-1}$, all three models agree that IC 750 has a Compton-thick line-of-sight column density to $>$ 99\% confidence. Compton-thick obscuration is well-documented to impinge substantial bias on the pursuit of SMBH AGN. Our results thus provide the first indication that Compton-thick obscuration should also be properly considered to uncover and understand the IMBH population in an unbiased manner., Comment: Accepted for publication in ApJ. 15 pages, 6 figures

Hard X-ray-selected samples of Active Galactic Nuclei (AGN) provide one of the cleanest views of supermassive black hole accretion, but are biased against objects obscured by Compton-thick gas column densities of $N_{\rm H}$ $>$ 10$^{24}$ cm$^{-2}$. To tackle this issue, we present the NuSTAR Local AGN $N_{\rm H}$ Distribution Survey (NuLANDS)$-$a legacy sample of 122 nearby ($z$ $<$ 0.044) AGN primarily selected to have warm infrared colors from IRAS between 25$-$60 $\mu$m. We show that optically classified type 1 and 2 AGN in NuLANDS are indistinguishable in terms of optical [OIII] line flux and mid-to-far infrared AGN continuum bolometric indicators, as expected from an isotropically selected AGN sample, while type 2 AGN are deficient in terms of their observed hard X-ray flux. By testing many X-ray spectroscopic models, we show the measured line-of-sight column density varies on average by $\sim$ 1.4 orders of magnitude depending on the obscurer geometry. To circumvent such issues we propagate the uncertainties per source into the parent column density distribution, finding a directly measured Compton-thick fraction of 35 $\pm$ 9%. By construction, our sample will miss sources affected by severe narrow-line reddening, and thus segregates sources dominated by small-scale nuclear obscuration from large-scale host-galaxy obscuration. This bias implies an even higher intrinsic obscured AGN fraction may be possible, although tests for additional biases arising from our infrared selection find no strong effects on the measured column-density distribution. NuLANDS thus holds potential as an optimized sample for future follow-up with current and next-generation instruments aiming to study the local AGN population in an isotropic manner., Comment: Accepted for publication in ApJ. 50 pages (78 including appendix and bibliography), 21 figures

The automatic evaluation of instruction following typically involves using large language models (LLMs) to assess response quality. However, there is a lack of comprehensive evaluation of these LLM-based evaluators across two dimensions: the base LLMs and the evaluation protocols. Therefore, we present a thorough meta-evaluation of instruction following, including 25 base LLMs and 15 recently proposed evaluation protocols, on 4 human-annotated datasets, assessing the evaluation accuracy of the LLM-evaluators. Our evaluation allows us to identify the best-performing base LLMs and evaluation protocols with a high degree of robustness. Moreover, our large-scale evaluation reveals: (1) Base LLM performance ranking remains largely consistent across evaluation protocols, with less capable LLMs showing greater improvement from protocol enhancements; (2) Robust evaluation of evaluation protocols requires many base LLMs with varying capability levels, as protocol effectiveness can depend on the base LLM used; (3) Evaluation results on different datasets are not always consistent, so a rigorous evaluation requires multiple datasets with distinctive features. We release our meta-evaluation suite ReIFE, which provides the codebase and evaluation result collection for more than 500 LLM-evaluator configurations, to support future research in instruction-following evaluation., Comment: GitHub Repo: https://github.com/yale-nlp/ReIFE, Evaluation Result Collection: https://huggingface.co/datasets/yale-nlp/ReIFE

The development of large language models (LLMs) has been instrumental in advancing state-of-the-art natural language processing applications. Training LLMs with billions of parameters and trillions of tokens require sophisticated distributed systems that enable composing and comparing several state-of-the-art techniques in order to efficiently scale across thousands of accelerators. However, existing solutions are complex, scattered across multiple libraries/repositories, lack interoperability, and are cumbersome to maintain. Thus, curating and empirically comparing training recipes require non-trivial engineering effort. This paper introduces TorchTitan, an open-source, PyTorch-native distributed training system that unifies state-of-the-art techniques, streamlining integration and reducing overhead. TorchTitan enables 3D parallelism in a modular manner with elastic scaling, providing comprehensive logging, checkpointing, and debugging tools for production-ready training. It also incorporates hardware-software co-designed solutions, leveraging features like Float8 training and SymmetricMemory. As a flexible test bed, TorchTitan facilitates custom recipe curation and comparison, allowing us to develop optimized training recipes for Llama 3.1 and provide guidance on selecting techniques for maximum efficiency based on our experiences. We thoroughly assess TorchTitan on the Llama 3.1 family of LLMs, spanning 8 billion to 405 billion parameters, and showcase its exceptional performance, modular composability, and elastic scalability. By stacking training optimizations, we demonstrate accelerations of 65.08% with 1D parallelism at the 128-GPU scale (Llama 3.1 8B), an additional 12.59% with 2D parallelism at the 256-GPU scale (Llama 3.1 70B), and an additional 30% with 3D parallelism at the 512-GPU scale (Llama 3.1 405B) on NVIDIA H100 GPUs over optimized baselines.

This white paper is the result of a collaboration by those that attended a workshop at the Facility for Rare Isotope Beams (FRIB), organized by the FRIB Theory Alliance (FRIB-TA), on Theoretical Justifications and Motivations for Early High-Profile FRIB Experiments. It covers a wide range of topics related to the science that will be explored at FRIB. After a brief introduction, the sections address: (II) Overview of theoretical methods, (III) Experimental capabilities, (IV) Structure, (V) Near-threshold Physics, (VI) Reaction mechanisms, (VII) Nuclear equations of state, (VIII) Nuclear astrophysics, (IX) Fundamental symmetries, and (X) Experimental design and uncertainty quantification., Comment: 227 pages, 24 figures

This research exploits the applications of reconfigurable intelligent surface (RIS)-assisted multiple input multiple output (MIMO) systems, specifically addressing the enhancement of communication reliability with modulated signals. Specifically, we first derive the analytical downlink symbol error rate (SER) of each user as a multivariate function of both the phase-shift and beamforming vectors. The analytical SER enables us to obtain insights into the synergistic dynamics between the RIS and MIMO communication. We then introduce a novel average SER minimization problem subject to the practical constraints of the transmitted power budget and phase shift coefficients, which is NP-hard. By incorporating the differential evolution (DE) algorithm as a pivotal tool for optimizing the intricate active and passive beamforming variables in RIS-assisted communication systems, the non-convexity of the considered SER optimization problem can be effectively handled. Furthermore, an efficient local search is incorporated into the DE algorithm to overcome the local optimum, and hence offer low SER and high communication reliability. Monte Carlo simulations validate the analytical results and the proposed optimization framework, indicating that the joint active and passive beamforming design is superior to the other benchmarks., Comment: 16 pages, 13 figures, and 1 table . Accepted by TWC

The Gamma-ray Transients Monitor (GTM) on board the Formosat-8B (FS-8B) satellite is designed to detect and localize Gamma-Ray Bursts (GRBs). By utilizing 2+2 CITIROC chips to manipulate 4+4 detectors, which are composed of GAGG(Ce) scintillators coupled with Silicon Photomultipliers (SiPMs) and oriented in various directions to achieve all-sky coverage, the GRB saturation fluences of GTM in the 50 keV to 1 MeV range for Short GRBs (SGRBs) and Long GRBs (LGRBs) were estimated to be about $3.1 \times 10^{-4}$ and $5.0 \times 10^{-3}\ {\rm erg/cm^2}$, respectively, based on simulations. To precisely interpret the GTM readout signal in terms of energy, several measurements for isotope and gain calibration were conducted. Despite encountering issues with crosstalk and SiPM saturation effect in the data, the energy spectrum can still be recovered by appropriately discarding channel noise and mapping with the correct ADC-to-energy relation. This paper summarizes the energy resolution of GTM and the linear variations in the relationship between photon energy and readout signal. At 662 keV, the energy resolution is about 16 %. Also, it demonstrates that greater gain is achieved by increasing voltage or decreasing temperature.

Open Radio Access Networks (O-RAN) are expected to revolutionize the telecommunications industry with benefits like cost reduction, vendor diversity, and improved network performance through AI optimization. Supporting the O-RAN ALLIANCE's mission to achieve more intelligent, open, virtualized and fully interoperable mobile networks, O-RAN Open Testing and Integration Centers (OTICs) play a key role in accelerating the adoption of O-RAN specifications based on rigorous testing and validation. One theme in the recent O-RAN Global PlugFest Spring 2024 focused on demonstrating consistent and repeatable Open Fronthaul testing in multiple labs. To respond to this topic, in this paper, we present a detailed analysis of the testing methodologies and results for O-RAN Distributed Unit (O-DU) in O-RAN across two OTICs. We identify key differences in testing setups, share challenges encountered, and propose best practices for achieving repeatable and consistent testing results. Our findings highlight the impact of different deployment technologies and testing environments on performance and conformance testing outcomes, providing valuable insights for future O-RAN implementations., Comment: Published version at RitiRAN Workshop - co-located with IEEE VTC Fall 2024

We present a comprehensive theoretical investigation of the strain-modulated excitonic properties of uni-axially strained transition-metal dichalcogenide monolayers (TMD-MLs) by solving the Bethe-Salpeter equation (BSE) established on the basis of first principles. We show that imposing an uni-axial strain onto a MoS_$2$ monolayers leads to the diversified band dispersions of the bright exciton (BX), gray exciton (GX), and dark exciton (DX) states, as a consequence of the competitive interplay between strain-induced valley drift (VD) and momentum-dependent electron-hole exchange interaction (EHEI). While the band dispersions of BX doublet in the light-accessible small reciprocal area remain almost unchanged against strain, the band dispersion of DX is reshaped by an increasing uni-axial strain from a parabola to a Mexican-hat-like profile, featured with unusual sign-reversal of the heavy effective mass and strain-activated brightness. In contrast, the effective mass of GX is drastically lightened by uni-axial strain and remains always positive. We show that the strain-diversified exciton band dispersions leads to the distinct exciton diffusivities and angle-resolved optical patterns of BX, GX, and DX in a strained TMD-ML, suggesting the feasibility of {\it spatially} resolving spinallowed and -forbidden excitons in exciton transport experiments and angle-resolved optical spectroscopies., Comment: 18pages, 5 figures

Reasoning is central to a wide range of intellectual activities, and while the capabilities of large language models (LLMs) continue to advance, their performance in reasoning tasks remains limited. The processes and mechanisms underlying reasoning are not yet fully understood, but key elements include path exploration, selection of relevant knowledge, and multi-step inference. Problems are solved through the synthesis of these components. In this paper, we propose a benchmark that focuses on a specific aspect of reasoning ability: the direct evaluation of multi-step inference. To this end, we design a special reasoning task where multi-step inference is specifically focused by largely eliminating path exploration and implicit knowledge utilization. Our dataset comprises pairs of explicit instructions and corresponding questions, where the procedures necessary for solving the questions are entirely detailed within the instructions. This setup allows models to solve problems solely by following the provided directives. By constructing problems that require varying numbers of steps to solve and evaluating responses at each step, we enable a thorough assessment of state-of-the-art LLMs' ability to follow instructions. To ensure the robustness of our evaluation, we include multiple distinct tasks. Furthermore, by comparing accuracy across tasks, utilizing step-aware metrics, and applying separately defined measures of complexity, we conduct experiments that offer insights into the capabilities and limitations of LLMs in reasoning tasks. Our findings have significant implications for the development of LLMs and highlight areas for future research in advancing their reasoning abilities. Our dataset is available at \url{https://huggingface.co/datasets/ifujisawa/procbench} and code at \url{https://github.com/ifujisawa/proc-bench}.

In personalized Federated Learning (pFL), high data heterogeneity can cause significant gradient divergence across devices, adversely affecting the learning process. This divergence, especially when gradients from different users form an obtuse angle during aggregation, can negate progress, leading to severe weight and gradient update degradation. To address this issue, we introduce a new approach to pFL design, namely Federated Learning with Layer-wise Aggregation via Gradient Analysis (FedLAG), utilizing the concept of gradient conflict at the layer level. Specifically, when layer-wise gradients of different clients form acute angles, those gradients align in the same direction, enabling updates across different clients toward identifying client-invariant features. Conversely, when layer-wise gradient pairs make create obtuse angles, the layers tend to focus on client-specific tasks. In hindsights, FedLAG assigns layers for personalization based on the extent of layer-wise gradient conflicts. Specifically, layers with gradient conflicts are excluded from the global aggregation process. The theoretical evaluation demonstrates that when integrated into other pFL baselines, FedLAG enhances pFL performance by a certain margin. Therefore, our proposed method achieves superior convergence behavior compared with other baselines. Extensive experiments show that our FedLAG outperforms several state-of-the-art methods and can be easily incorporated with many existing methods to further enhance performance.

Tidal disruption events (TDEs) could be an important growth channel for massive black holes in dwarf galaxies. Theoretical work suggests that the observed active galactic nuclei (AGNs) in dwarf galaxies are predominantly TDE-powered. To assess this claim, we perform variability analyses on the dwarf-hosted AGNs detected in the $7$ Ms Chandra Deep Field-South (CDF-S) survey, with observations spanning $\approx 16$ years. Based on the spectral energy distribution (SED) modeling with X-CIGALE, we select AGNs hosted by dwarf galaxies (stellar mass below $10^{10}\ M_\odot$). We focus on X-ray sources with full-band detections, leading to a sample of $78$ AGNs (0.122 $\leq$ $z$ $\leq$ 3.515). We fit the X-ray light curves with a canonical TDE model of $t^{-5/3}$ and a constant model. If the former outperforms the latter in fitting quality for a source, we consider the source as a potential TDE. We identify five potential TDEs, constituting a small fraction of our sample. Using true- and false-positive rates obtained from fitting models to simulated light curves, we perform Bayesian analysis to obtain the posterior of the TDE fraction for our sample. The posterior peaks close to zero ($2.56\%$), and we obtain a $2$-$\sigma$ upper limit of $9.80\%$. Therefore, our result indicates that the observed AGNs in dwarf galaxies are not predominantly powered by TDEs., Comment: 16 pages, 10 figures, accepted for publication in ApJ

Magnetic reconnection is a ubiquitous fundamental process in space and astrophysical plasmas that rapidly converts magnetic energy into some combination of flow energy, thermal energy, and non-thermal energetic particles. Over the past decade, a new experimental platform has been developed to study magnetic reconnection using strong coil currents powered by high power lasers at low plasma beta, typical conditions under which reconnection is energetically important in astrophysics. KJ-class lasers were used to drive parallel currents to reconnect MG-level magnetic fields in a quasi-axisymmetric geometry, similar to the Magnetic Reconnection Experiment or MRX, and thus this platform is named micro-MRX. This presentation summarizes two major findings from micro-MRX: direct measurement of accelerated electrons and observation of ion acoustic waves during anti-parallel reconnection. The angular dependence of the measured electron energy spectrum and the resulting accelerated energies, supported by particle-in-cell simulations, indicate that direct acceleration by the out-of-plane reconnection electric field is at work. Furthermore, a sudden onset of ion acoustic bursts has been measured by collective Thomson scattering in the exhaust of magnetic reconnection, followed by electron acoustic bursts with electron heating and bulk acceleration. These results demonstrate that the micro-MRX platform offers a novel and unique approach to study magnetic reconnection in the laboratory in addition to the capabilities provided by traditional magnetized plasma experiments such as MRX and the upcoming FLARE (Facility for Laboratory Reconnection experiments). Future approaches to study other particle acceleration mechanisms and ion acoustic waves from magnetic reconnection are also discussed., Comment: 16 pages, 13 figures, 89 references, accepted for publication in Physics of Plasmas

Post-training Large Language Models (LLMs) with explicit reasoning trajectories can enhance their reasoning abilities. However, acquiring such high-quality trajectory data typically demands meticulous supervision from humans or superior models, which can be either expensive or license-constrained. In this paper, we explore how far an LLM can improve its reasoning by self-synthesizing reasoning paths as training data without any additional supervision. Existing self-synthesizing methods, such as STaR, suffer from poor generalization to out-of-domain (OOD) reasoning tasks. We hypothesize it is due to that their self-synthesized reasoning paths are too task-specific, lacking general task-agnostic reasoning guidance. To address this, we propose Reasoning Generalist via Self-Improvement (ReGenesis), a method to self-synthesize reasoning paths as post-training data by progressing from abstract to concrete. More specifically, ReGenesis self-synthesizes reasoning paths by converting general reasoning guidelines into task-specific ones, generating reasoning structures, and subsequently transforming these structures into reasoning paths, without the need for human-designed task-specific examples used in existing methods. We show that ReGenesis achieves superior performance on all in-domain and OOD settings tested compared to existing methods. For six OOD tasks specifically, while previous methods exhibited an average performance decrease of approximately 4.6% after post training, ReGenesis delivers around 6.1% performance improvement. We also conduct in-depth analysis of our framework and show ReGenesis is effective across various LLMs and design choices.

We perform first-principles electron-phonon calculations to evaluate the superconducting transition temperature $T_c$ for ternary superhard metals B$_2$C$_3$N and B$_4$C$_5$N$_3$. These materials are predicted to exhibit a wide distribution of electron-phonon coupling parameters on their multiple Fermi surface sheets, which necessitates solving the anisotropic Eliashberg equations for accurate determination of $T_c$. An ambient-pressure $T_c$ of $\sim 40$ K and $\sim 20$ K is obtained respectively for B$_2$C$_3$N and B$_4$C$_5$N$_3$ from the anisotropic gap equations. The relatively high $T_c$ of these compounds is due in part to their high Debye temperatures associated with superhardness. The materials under study are potentially synthesizable, as their formation energies are comparable to those of other recently synthesized superhard B-C-N compounds. Therefore, studying superhard metals could hold the promise of realizing new higher-$T_c$ superconductors at ambient pressure., Comment: 7 pages, 4 figures

We study the Differential Privacy (DP) guarantee of hidden-state Noisy-SGD algorithms over a bounded domain. Standard privacy analysis for Noisy-SGD assumes all internal states are revealed, which leads to a divergent R'enyi DP bound with respect to the number of iterations. Ye & Shokri (2022) and Altschuler & Talwar (2022) proved convergent bounds for smooth (strongly) convex losses, and raise open questions about whether these assumptions can be relaxed. We provide positive answers by proving convergent R'enyi DP bound for non-convex non-smooth losses, where we show that requiring losses to have H\"older continuous gradient is sufficient. We also provide a strictly better privacy bound compared to state-of-the-art results for smooth strongly convex losses. Our analysis relies on the improvement of shifted divergence analysis in multiple aspects, including forward Wasserstein distance tracking, identifying the optimal shifts allocation, and the H"older reduction lemma. Our results further elucidate the benefit of hidden-state analysis for DP and its applicability.