Search

Asian Americans as a group are overrepresented among STEM college graduates and have the highest average college enrollment rate of any racial or ethnic category. Thus, Asian Americans are typically excluded from educational interventions directed at improving STEM education for Students of Color because they are not considered to be underrepresented minorities. However, statistics obscure the individual needs of the more than 20 ethnic subgroups that fall under the umbrella term Asian Americans. Using a participatory action research approach, this paper documents the institutional and sociocultural factors that push out HMoob (or Hmong) American college students from STEM programs at one large, predominantly White university; and the coordinate processes of gatekeeping and transactional advising that either redirect those students toward non-STEM programs or force them out of the university completely.

We introduce our first-generation reasoning models, DeepSeek-R1-Zero and DeepSeek-R1. DeepSeek-R1-Zero, a model trained via large-scale reinforcement learning (RL) without supervised fine-tuning (SFT) as a preliminary step, demonstrates remarkable reasoning capabilities. Through RL, DeepSeek-R1-Zero naturally emerges with numerous powerful and intriguing reasoning behaviors. However, it encounters challenges such as poor readability, and language mixing. To address these issues and further enhance reasoning performance, we introduce DeepSeek-R1, which incorporates multi-stage training and cold-start data before RL. DeepSeek-R1 achieves performance comparable to OpenAI-o1-1217 on reasoning tasks. To support the research community, we open-source DeepSeek-R1-Zero, DeepSeek-R1, and six dense models (1.5B, 7B, 8B, 14B, 32B, 70B) distilled from DeepSeek-R1 based on Qwen and Llama.

We present DeepSeek-V3, a strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token. To achieve efficient inference and cost-effective training, DeepSeek-V3 adopts Multi-head Latent Attention (MLA) and DeepSeekMoE architectures, which were thoroughly validated in DeepSeek-V2. Furthermore, DeepSeek-V3 pioneers an auxiliary-loss-free strategy for load balancing and sets a multi-token prediction training objective for stronger performance. We pre-train DeepSeek-V3 on 14.8 trillion diverse and high-quality tokens, followed by Supervised Fine-Tuning and Reinforcement Learning stages to fully harness its capabilities. Comprehensive evaluations reveal that DeepSeek-V3 outperforms other open-source models and achieves performance comparable to leading closed-source models. Despite its excellent performance, DeepSeek-V3 requires only 2.788M H800 GPU hours for its full training. In addition, its training process is remarkably stable. Throughout the entire training process, we did not experience any irrecoverable loss spikes or perform any rollbacks. The model checkpoints are available at https://github.com/deepseek-ai/DeepSeek-V3.

The vision of a broadly capable and goal-directed agent, such as an Internet-browsing agent in the digital world and a household humanoid in the physical world, has rapidly advanced, thanks to the generalization capability of foundation models. Such a generalist agent needs to have a large and diverse skill repertoire, such as finding directions between two travel locations and buying specific items from the Internet. If each skill needs to be specified manually through a fixed set of human-annotated instructions, the agent's skill repertoire will necessarily be limited due to the quantity and diversity of human-annotated instructions. In this work, we address this challenge by proposing Proposer-Agent-Evaluator, an effective learning system that enables foundation model agents to autonomously discover and practice skills in the wild. At the heart of PAE is a context-aware task proposer that autonomously proposes tasks for the agent to practice with context information of the environment such as user demos or even just the name of the website itself for Internet-browsing agents. Then, the agent policy attempts those tasks with thoughts and actual grounded operations in the real world with resulting trajectories evaluated by an autonomous VLM-based success evaluator. The success evaluation serves as the reward signal for the agent to refine its policies through RL. We validate PAE on challenging vision-based web navigation, using both real-world and self-hosted websites from WebVoyager and WebArena.To the best of our knowledge, this work represents the first effective learning system to apply autonomous task proposal with RL for agents that generalizes real-world human-annotated benchmarks with SOTA performances. Our open-source checkpoints and code can be found in https://yanqval.github.io/PAE/

The rapid development of next-generation networking technologies underscores their transformative role in revolutionizing modern communication systems, enabling faster, more reliable, and highly interconnected solutions. However, such development has also brought challenges to network optimizations. Thanks to the emergence of Large Language Models (LLMs) in recent years, tools including Retrieval Augmented Generation (RAG) have been developed and applied in various fields including networking, and have shown their effectiveness. Taking one step further, the integration of knowledge graphs into RAG frameworks further enhanced the performance of RAG in networking applications such as Intent-Driven Networks (IDNs) and spectrum knowledge maps by providing more contextually relevant responses through more accurate retrieval of related network information. This paper introduces the RAG framework that integrates knowledge graphs in its database and explores such framework's application in networking. We begin by exploring RAG's applications in networking and the limitations of conventional RAG and present the advantages that knowledge graphs' structured knowledge representation brings to the retrieval and generation processes. Next, we propose a detailed GraphRAG-based framework for networking, including a step-by-step tutorial on its construction. Our evaluation through a case study on channel gain prediction demonstrates GraphRAG's enhanced capability in generating accurate, contextually rich responses, surpassing traditional RAG models. Finally, we discuss key future directions for applying knowledge-graphs-empowered RAG frameworks in networking, including robust updates, mitigation of hallucination, and enhanced security measures for networking applications., Comment: 9 pages, 4 figures

This article clarifies historical and sociocultural factors that impact the role of STEM in the racialization of Asian Americans. Drawing on critical race and other theories of Asian American racialization, and a review of empirical research on the experiences of Asian American college students in STEM, we develop a conceptual framework called "STEM Asianization" that highlights the role of STEM ideology in the model minority racialization of Asian Americans. Consequences for Asian American students include (1) erasure of the intersectional experiences of minoritized Asian American students; (2) dehumanization of Asian Americans and establishment of a bamboo ceiling; (3) representation of Asian Americans as a perpetual foreigner/Yellow Peril during times of cultural and political crisis; and (4) representation of Asian Americans who cannot or do not conform to the STEM achievement narrative as a failed minority. We argue that STEM Asianization reproduces White supremacy by ideologically reinforcing the colorblind meritocracy of STEM institutions in the United States. [Additional funding provided by the Wisconsin Louis Stokes Alliance for Minority Participation.]

Large language models (LLMs) have demonstrated immense potential across various tasks. However, research for exploring and improving the capabilities of LLMs in interpreting graph structures remains limited. To address this gap, we conduct a comprehensive evaluation of prompting current open-source LLMs on graph-to-text generation tasks. Although we explored the optimal prompting strategies and proposed a novel and effective diversity-difficulty-based few-shot sample selection method, we found that the improvements from tuning-free approaches were incremental, as LLMs struggle with planning on complex graphs, particularly those with a larger number of triplets. To further improve LLMs in planning with graph sequences and grounding in truth, we introduce a new graph-to-text dataset, PlanGTG, annotated with two sub-tasks: reordering and attribution. Through extensive automatic and human evaluations, we demonstrate significant improvements in the quality of generated text from both few-shot learning and fine-tuning perspectives using the PlanGTG dataset. Our study paves the way for new research directions in graph-to-text generation. PlanGTG datasets can be found in https://github.com/probe2/kg_text., Comment: NAACL 2025

Multi-modal sequential recommendation (SR) leverages multi-modal data to learn more comprehensive item features and user preferences than traditional SR methods, which has become a critical topic in both academia and industry. Existing methods typically focus on enhancing multi-modal information utility through adaptive modality fusion to capture the evolving of user preference from user-item interaction sequences. However, most of them overlook the interference caused by redundant interest-irrelevant information contained in rich multi-modal data. Additionally, they primarily rely on implicit temporal information based solely on chronological ordering, neglecting explicit temporal signals that could more effectively represent dynamic user interest over time. To address these limitations, we propose a Hierarchical time-aware Mixture of experts for multi-modal Sequential Recommendation (HM4SR) with a two-level Mixture of Experts (MoE) and a multi-task learning strategy. Specifically, the first MoE, named Interactive MoE, extracts essential user interest-related information from the multi-modal data of each item. Then, the second MoE, termed Temporal MoE, captures user dynamic interests by introducing explicit temporal embeddings from timestamps in modality encoding. To further address data sparsity, we propose three auxiliary supervision tasks: sequence-level category prediction (CP) for item feature understanding, contrastive learning on ID (IDCL) to align sequence context with user interests, and placeholder contrastive learning (PCL) to integrate temporal information with modalities for dynamic interest modeling. Extensive experiments on four public datasets verify the effectiveness of HM4SR compared to several state-of-the-art approaches., Comment: Accepted to WWW 2025

Recent advances in Large Language Models (LLMs) have enabled the development of Video-LLMs, advancing multimodal learning by bridging video data with language tasks. However, current video understanding models struggle with processing long video sequences, supporting multi-turn dialogues, and adapting to real-world dynamic scenarios. To address these issues, we propose StreamChat, a training-free framework for streaming video reasoning and conversational interaction. $\StreamChat$ leverages a novel hierarchical memory system to efficiently process and compress video features over extended sequences, enabling real-time, multi-turn dialogue. Our framework incorporates a parallel system scheduling strategy that enhances processing speed and reduces latency, ensuring robust performance in real-world applications. Furthermore, we introduce StreamBench, a versatile benchmark that evaluates streaming video understanding across diverse media types and interactive scenarios, including multi-turn interactions and complex reasoning tasks. Extensive evaluations on StreamBench and other public benchmarks demonstrate that StreamChat significantly outperforms existing state-of-the-art models in terms of accuracy and response times, confirming its effectiveness for streaming video understanding. Code is available at StreamChat: https://github.com/hmxiong/StreamChat., Comment: Accepted to ICLR 2025. Code is available at https://github.com/hmxiong/StreamChat

Observability in cloud infrastructure is critical for service providers, driving the widespread adoption of anomaly detection systems for monitoring metrics. However, existing systems often struggle to simultaneously achieve explainability, reproducibility, and autonomy, which are three indispensable properties for production use. We introduce Argos, an agentic system for detecting time-series anomalies in cloud infrastructure by leveraging large language models (LLMs). Argos proposes to use explainable and reproducible anomaly rules as intermediate representation and employs LLMs to autonomously generate such rules. The system will efficiently train error-free and accuracy-guaranteed anomaly rules through multiple collaborative agents and deploy the trained rules for low-cost online anomaly detection. Through evaluation results, we demonstrate that Argos outperforms state-of-the-art methods, increasing $F_1$ scores by up to $9.5\%$ and $28.3\%$ on public anomaly detection datasets and an internal dataset collected from Microsoft, respectively.

In this paper, we explore a novel federated multimodal instruction tuning task(FedMIT), which is significant for collaboratively fine-tuning MLLMs on different types of multimodal instruction data on distributed devices. To solve the new task, we propose a federated multimodal instruction tuning framework(Pilot). Our framework integrates two stages of "adapter on adapter" into the connector of the vision encoder and the LLM. In stage 1, we extract task-specific features and client-specific features from visual information. In stage 2, we build the cross-task Mixture-of-Adapters(CT-MoA) module to perform cross-task interaction. Each client can not only capture personalized information of local data and learn task-related multimodal information, but also learn general knowledge from other tasks. In addition, we introduce an adaptive parameter aggregation strategy for text training parameters, which optimizes parameter aggregation by calculating weights based on the euclidean distance between parameters, so that parameter aggregation can benefit from positive effects to the greatest extent while effectively reducing negative effects. Our framework can collaboratively exploit distributed data from different local clients to learn cross-task knowledge without being affected by the task heterogeneity during instruction tuning. The effectiveness of our method is verified in two different cross-task scenarios.

We investigate the optical properties of mini boson stars within the framework of Palatini $f(R)$ gravity, adopting a quadratic form $f(R) = R + \xi R^2$, where $\xi$ is the gravitational coupling constant. By deriving the modified scalar Lagrangian and solving the field equations numerically, we explore photon trajectories and the resulting optical images under spherical light sources and thin accretion disks. Unlike Schwarzschild black holes (BHs), boson stars lack stable photon rings due to the positive second derivative of their effective potential. Consequently, their images are dominated by direct emissions from photons completing a single orbit. The study examines the dependence of the optical characteristics on the initial scalar field $\psi_0$ and the coupling parameter $\xi$. Numerical results include effective potentials, redshift maps, and detailed imaging of boson stars, providing insights into distinguishing boson stars from black holes using high-resolution astronomical observations., Comment: 29 pages, 14 figures

Learned image compression (LIC) has demonstrated superior rate-distortion (R-D) performance compared to traditional codecs, but is challenged by training inefficiency that could incur more than two weeks to train a state-of-the-art model from scratch. Existing LIC methods overlook the slow convergence caused by compacting energy in learning nonlinear transforms. In this paper, we first reveal that such energy compaction consists of two components, i.e., feature decorrelation and uneven energy modulation. On such basis, we propose a linear auxiliary transform (AuxT) to disentangle energy compaction in training nonlinear transforms. The proposed AuxT obtains coarse approximation to achieve efficient energy compaction such that distribution fitting with the nonlinear transforms can be simplified to fine details. We then develop wavelet-based linear shortcuts (WLSs) for AuxT that leverages wavelet-based downsampling and orthogonal linear projection for feature decorrelation and subband-aware scaling for uneven energy modulation. AuxT is lightweight and plug-and-play to be integrated into diverse LIC models to address the slow convergence issue. Experimental results demonstrate that the proposed approach can accelerate training of LIC models by 2 times and simultaneously achieves an average 1\% BD-rate reduction. To our best knowledge, this is one of the first successful attempt that can significantly improve the convergence of LIC with comparable or superior rate-distortion performance. Code will be released at \url{https://github.com/qingshi9974/AuxT}, Comment: Accepted by ICLR2025

The paper introduces EICopilot, an novel agent-based solution enhancing search and exploration of enterprise registration data within extensive online knowledge graphs like those detailing legal entities, registered capital, and major shareholders. Traditional methods necessitate text-based queries and manual subgraph explorations, often resulting in time-consuming processes. EICopilot, deployed as a chatbot via Baidu Enterprise Search, improves this landscape by utilizing Large Language Models (LLMs) to interpret natural language queries. This solution automatically generates and executes Gremlin scripts, providing efficient summaries of complex enterprise relationships. Distinct feature a data pre-processing pipeline that compiles and annotates representative queries into a vector database of examples for In-context learning (ICL), a comprehensive reasoning pipeline combining Chain-of-Thought with ICL to enhance Gremlin script generation for knowledge graph search and exploration, and a novel query masking strategy that improves intent recognition for heightened script accuracy. Empirical evaluations demonstrate the superior performance of EICopilot, including speed and accuracy, over baseline methods, with the \emph{Full Mask} variant achieving a syntax error rate reduction to as low as 10.00% and an execution correctness of up to 82.14%. These components collectively contribute to superior querying capabilities and summarization of intricate datasets, positioning EICopilot as a groundbreaking tool in the exploration and exploitation of large-scale knowledge graphs for enterprise information search.

The event-based Vision-Language Model (VLM) recently has made good progress for practical vision tasks. However, most of these works just utilize CLIP for focusing on traditional perception tasks, which obstruct model understanding explicitly the sufficient semantics and context from event streams. To address the deficiency, we propose EventVL, the first generative event-based MLLM (Multimodal Large Language Model) framework for explicit semantic understanding. Specifically, to bridge the data gap for connecting different modalities semantics, we first annotate a large event-image/video-text dataset, containing almost 1.4 million high-quality pairs of data, which enables effective learning across various scenes, e.g., drive scene or human motion. After that, we design Event Spatiotemporal Representation to fully explore the comprehensive information by diversely aggregating and segmenting the event stream. To further promote a compact semantic space, Dynamic Semantic Alignment is introduced to improve and complete sparse semantic spaces of events. Extensive experiments show that our EventVL can significantly surpass existing MLLM baselines in event captioning and scene description generation tasks. We hope our research could contribute to the development of the event vision community.

State Space Models (SSMs), as key components of Mamaba, have gained increasing attention for vision models recently, thanks to their efficient long sequence modeling capability. Given the computational cost of deploying SSMs on resource-limited edge devices, Post-Training Quantization (PTQ) is a technique with the potential for efficient deployment of SSMs. In this work, we propose QMamba, one of the first PTQ frameworks to our knowledge, designed for vision SSMs based on the analysis of the activation distributions in SSMs. We reveal that the distribution of discrete parameters exhibits long-tailed skewness and the distribution of the hidden state sequence exhibits highly dynamic variations. Correspondingly, we design Long-tailed Skewness Quantization (LtSQ) to quantize discrete parameters and Temporal Group Quantization (TGQ) to quantize hidden states, which reduces the quantization errors. Extensive experiments demonstrate that QMamba outperforms advanced PTQ methods on vision models across multiple model sizes and architectures. Notably, QMamba surpasses existing methods by 21.0% on ImageNet classification with 4-bit activations.

Synthetic Aperture Radar (SAR) stands as an indispensable sensor for Earth observation, owing to its unique capability for all-day imaging. Nevertheless, in a data-driven era, the scarcity of large-scale datasets poses a significant bottleneck to advancing SAR automatic target recognition (ATR) technology. This paper introduces NUDT4MSTAR, a large-scale SAR dataset for vehicle target recognition in the wild, including 40 target types and a wide array of imaging conditions across 5 different scenes. NUDT4MSTAR represents a significant leap forward in dataset scale, containing over 190,000 images-tenfold the size of its predecessors. To enhance the utility of this dataset, we meticulously annotate each image with detailed target information and imaging conditions. We also provide data in both processed magnitude images and original complex formats. Then, we construct a comprehensive benchmark consisting of 7 experiments with 15 recognition methods focusing on the stable and effective ATR issues. Besides, we conduct transfer learning experiments utilizing various models trained on NUDT4MSTAR and applied to three other target datasets, thereby demonstrating its substantial potential to the broader field of ground objects ATR. Finally, we discuss this dataset's application value and ATR's significant challenges. To the best of our knowledge, this work marks the first-ever endeavor to create a large-scale dataset benchmark for fine-grained SAR recognition in the wild, featuring an extensive collection of exhaustively annotated vehicle images. We expect that the open source of NUDT4MSTAR will facilitate the development of SAR ATR and attract a wider community of researchers., Comment: 18 pages, 15 figures; link: https://github.com/waterdisappear/NUDT4MSTAR

Generating high-quality textures for 3D scenes is crucial for applications in interior design, gaming, and augmented/virtual reality (AR/VR). Although recent advancements in 3D generative models have enhanced content creation, significant challenges remain in achieving broad generalization and maintaining style consistency across multiple viewpoints. Current methods, such as 2D diffusion models adapted for 3D texturing, suffer from lengthy processing times and visual artifacts, while approaches driven by 3D data often fail to generalize effectively. To overcome these challenges, we introduce InsTex, a two-stage architecture designed to generate high-quality, style-consistent textures for 3D indoor scenes. InsTex utilizes depth-to-image diffusion priors in a coarse-to-fine pipeline, first generating multi-view images with a pre-trained 2D diffusion model and subsequently refining the textures for consistency. Our method supports both textual and visual prompts, achieving state-of-the-art results in visual quality and quantitative metrics, and demonstrates its effectiveness across various 3D texturing applications.

Detecting anomalies in temporal data has gained significant attention across various real-world applications, aiming to identify unusual events and mitigate potential hazards. In practice, situations often involve a mix of segment-level labels (detected abnormal events with segments of time points) and unlabeled data (undetected events), while the ideal algorithmic outcome should be point-level predictions. Therefore, the huge label information gap between training data and targets makes the task challenging. In this study, we formulate the above imperfect information as noisy labels and propose NRdetector, a noise-resilient framework that incorporates confidence-based sample selection, robust segment-level learning, and data-centric point-level detection for multivariate time series anomaly detection. Particularly, to bridge the information gap between noisy segment-level labels and missing point-level labels, we develop a novel loss function that can effectively mitigate the label noise and consider the temporal features. It encourages the smoothness of consecutive points and the separability of points from segments with different labels. Extensive experiments on real-world multivariate time series datasets with 11 different evaluation metrics demonstrate that NRdetector consistently achieves robust results across multiple real-world datasets, outperforming various baselines adapted to operate in our setting., Comment: Accepted by 2025 ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD'25)

Language model pretraining with next token prediction has proved effective for scaling compute but is limited to the amount of available training data. Scaling reinforcement learning (RL) unlocks a new axis for the continued improvement of artificial intelligence, with the promise that large language models (LLMs) can scale their training data by learning to explore with rewards. However, prior published work has not produced competitive results. In light of this, we report on the training practice of Kimi k1.5, our latest multi-modal LLM trained with RL, including its RL training techniques, multi-modal data recipes, and infrastructure optimization. Long context scaling and improved policy optimization methods are key ingredients of our approach, which establishes a simplistic, effective RL framework without relying on more complex techniques such as Monte Carlo tree search, value functions, and process reward models. Notably, our system achieves state-of-the-art reasoning performance across multiple benchmarks and modalities -- e.g., 77.5 on AIME, 96.2 on MATH 500, 94-th percentile on Codeforces, 74.9 on MathVista -- matching OpenAI's o1. Moreover, we present effective long2short methods that use long-CoT techniques to improve short-CoT models, yielding state-of-the-art short-CoT reasoning results -- e.g., 60.8 on AIME, 94.6 on MATH500, 47.3 on LiveCodeBench -- outperforming existing short-CoT models such as GPT-4o and Claude Sonnet 3.5 by a large margin (up to +550%)., Comment: 25 pages

The deployment of LoRa networks necessitates joint performance optimization, including packet delivery rate, energy efficiency, and throughput. Additionally, multiple LoRa parameters for packet transmission must be dynamically configured to tailor the performance metrics prioritization across varying channel environments. Because of the coupling relationship between LoRa parameters and metrics, existing works have opted to focus on certain parameters or specific metrics to circumvent the intricate coupling relationship, leading to limited adaptability. Therefore, we propose D-LoRa, a distributed parameter adaptation scheme, based on reinforcement learning towards network performance. We decompose the joint performance optimization problem into multiple independent Multi-Armed Bandit (MAB) problems with different reward functions. We have also built a comprehensive analytical model for the LoRa network that considers path loss, quasi-orthogonality of spreading factor, and packet collision. Experimental results show that our scheme can increase packet delivery rate by up to 28.8% and demonstrates superior adaptability across different performance metrics.

Embedding quantum dot circuits into microwave cavities has emerged as a novel platform for controlling photon emission statistics by electrical means. With such a model, we reveal previously undefined quantum phase transitions in electrically driven lasing regimes by breaking the photon gain-loss balance condition. For one-photon interaction, the scaling theory indicates that the system undergoes a continuous phase transition from thermal to coherent photon emissions, consistent with conventional laser physics. Going beyond this, a hidden discontinuous quantum phase transition from superbunched to coherent states in two-photon processes, accompanied by the bistability within a mean-field theory, is predicted. Our prediction, along with its extension to multiphoton processes, represents a key step towards accessing lasing phase transitions.

Tabular data is one of the most widely used formats across industries, driving critical applications in areas such as finance, healthcare, and marketing. In the era of data-centric AI, improving data quality and representation has become essential for enhancing model performance, particularly in applications centered around tabular data. This survey examines the key aspects of tabular data-centric AI, emphasizing feature selection and feature generation as essential techniques for data space refinement. We provide a systematic review of feature selection methods, which identify and retain the most relevant data attributes, and feature generation approaches, which create new features to simplify the capture of complex data patterns. This survey offers a comprehensive overview of current methodologies through an analysis of recent advancements, practical applications, and the strengths and limitations of these techniques. Finally, we outline open challenges and suggest future perspectives to inspire continued innovation in this field.

Macroscopic entanglement, as critical quantum resources in quantum information science, has been extensively studied in optomechanical systems with purely dispersive coupling over the past decades. However, quantum entanglement, induced by purely dissipative coupling, remains unexplored. In this work, we study quantum entanglement in a Michelson-Sagnac interferometer, where the dispersive and the dissipative coupling can be arbitrarily switched on and off. With current experimental parameters, we demonstrate that the steady-state mechanical displacement exhibits a nonlinear (linear) dependence on the driving power with the purely dispersive (dissipative) coupling. Further, we find that quantum entanglement generated by purely dissipative coupling is significantly stronger and more noise-tolerant than that generated by purely dispersive coupling. When both couplings coexist, entanglement is weakened due to interference. Our findings provide a promising path to engineer strong and noise-tolerant quantum entanglement in purely dissipative quantum systems., Comment: 7 pages, 5 figures

Networked cybernetic and physical systems of the Internet of Things (IoT) immerse civilian and industrial infrastructures into an interconnected and dynamic web of hybrid and mobile devices. The key feature of such systems is the hybrid and tight coupling of mobile and pervasive discrete communications in a continuously evolving environment (discrete computations with predominant continuous dynamics). In the aim of ensuring the correctness and reliability of such heterogeneous infrastructures, we introduce the hybrid {\pi}-calculus (HpC), to formally capture both mobility, pervasiveness and hybridisation in infrastructures where the network topology and its communicating entities evolve continuously in the physical world. The {\pi}-calculus proposed by Robin Milner et al. is a process calculus that can model mobile communications and computations in a very elegant manner. The HpC we propose is a conservative extension of the classical {\pi}-calculus, i.e., the extension is ``minimal'', and yet describes mobility, time and physics of systems, while allowing to lift all theoretical results (e.g. bisimulation) to the context of that extension. We showcase the HpC by considering a realistic handover protocol among mobile devices., Comment: The published version of this article will be available in the ACM Digital Library as part of the Proceedings of the ACM on Programming Languages issue for SPLASH/OOPSLA 2025. This extended version contains additional appendices, proofs and case studies

Mobile edge Large Language Model (LLM) deployments face inherent constraints, such as limited computational resources and network bandwidth. Although Retrieval-Augmented Generation (RAG) mitigates some challenges by integrating external knowledge bases, inefficient cache management can still result in high retrieval latency and frequent cache updates. To address these issues, we propose an Adaptive Contextual Caching (ACC) framework that anticipates user needs by proactively caching semantically relevant data for mobile-edge LLMs. ACC utilizes a deep reinforcement learning (DRL) module to refine cache replacement policies, balancing user context, document similarity, and the overhead associated with cache misses. Experimental results demonstrate that ACC increases cache hit rates to over 80\% after only 11 training episodes, outperforming FIFO, LRU, and semantic-only caching while reducing retrieval latency by up to 40\%. In particular, ACC also reduces local caching overhead (i.e., the cost of updating the cache when a miss occurs) by as much as 55\%, enabling scalable, low-latency LLM services in resource-constrained edge environments., Comment: 8 pages, 5 figures

Single image defocus deblurring (SIDD) aims to restore an all-in-focus image from a defocused one. Distribution shifts in defocused images generally lead to performance degradation of existing methods during out-of-distribution inferences. In this work, we gauge the intrinsic reason behind the performance degradation, which is identified as the heterogeneity of lens-specific point spread functions. Empirical evidence supports this finding, motivating us to employ a continual test-time adaptation (CTTA) paradigm for SIDD. However, traditional CTTA methods, which primarily rely on entropy minimization, cannot sufficiently explore task-dependent information for pixel-level regression tasks like SIDD. To address this issue, we propose a novel Siamese networks-based continual test-time adaptation framework, which adapts source models to continuously changing target domains only requiring unlabeled target data in an online manner. To further mitigate semantically erroneous textures introduced by source SIDD models under severe degradation, we revisit the learning paradigm through a structural causal model and propose Causal Siamese networks (CauSiam). Our method leverages large-scale pre-trained vision-language models to derive discriminative universal semantic priors and integrates these priors into Siamese networks, ensuring causal identifiability between blurry inputs and restored images. Extensive experiments demonstrate that CauSiam effectively improves the generalization performance of existing SIDD methods in continuously changing domains.

Vision Transformers (ViTs) are increasingly being adopted in various sensitive vision applications - like medical diagnosis, facial recognition, etc. To improve the interpretability of such models, many approaches attempt to forward-align them with carefully annotated abstract, human-understandable semantic entities - concepts. Concepts provide global rationales to the model predictions and can be quickly understood/intervened on by domain experts. Most current research focuses on designing model-agnostic, plug-and-play generic concept-based explainability modules that do not incorporate the inner workings of foundation models (e.g., inductive biases, scale invariance, etc.) during training. To alleviate this issue for ViTs, in this paper, we propose a novel Concept Representation Alignment Module (CRAM) which learns both scale and position-aware representations from multi-scale feature pyramids and patch representations respectively. CRAM further aligns these representations with concept annotations through an attention matrix. The proposed CRAM module improves the predictive performance of ViT architectures and also provides accurate and robust concept explanations as demonstrated on five datasets - including three widely used benchmarks (CUB, Pascal APY, Concept-MNIST) and 2 real-world datasets (AWA2, KITS).

Reinforcement Learning (RL) has shown excellent performance in solving decision-making and control problems of autonomous driving, which is increasingly applied in diverse driving scenarios. However, driving is a multi-attribute problem, leading to challenges in achieving multi-objective compatibility for current RL methods, especially in both policy execution and policy iteration. On the one hand, the common action space structure with single action type limits driving flexibility or results in large behavior fluctuations during policy execution. On the other hand, the multi-attribute weighted single reward function result in the agent's disproportionate attention to certain objectives during policy iterations. To this end, we propose a Multi-objective Ensemble-Critic reinforcement learning method with Hybrid Parametrized Action for multi-objective compatible autonomous driving. Specifically, a parameterized action space is constructed to generate hybrid driving actions, combining both abstract guidance and concrete control commands. A multi-objective critics architecture is constructed considering multiple attribute rewards, to ensure simultaneously focusing on different driving objectives. Additionally, uncertainty-based exploration strategy is introduced to help the agent faster approach viable driving policy. The experimental results in both the simulated traffic environment and the HighD dataset demonstrate that our method can achieve multi-objective compatible autonomous driving in terms of driving efficiency, action consistency, and safety. It enhances the general performance of the driving while significantly increasing training efficiency., Comment: 12 pages, 9 figures, 5 tables

Traditional Mixture-of-Experts (MoE) networks benefit from utilizing multiple smaller expert models as opposed to a single large network. However, these experts typically operate independently, leaving a question open about whether interconnecting these models could enhance the performance of MoE networks. In response, we introduce GRAPHMOE, a novel method aimed at augmenting the cognitive depth of language models via a self-rethinking mechanism constructed on Pseudo GraphMoE networks. GRAPHMOE employs a recurrent routing strategy to simulate iterative thinking steps, thereby facilitating the flow of information among expert nodes. We implement the GRAPHMOE architecture using Low-Rank Adaptation techniques (LoRA) and conduct extensive experiments on various benchmark datasets. The experimental results reveal that GRAPHMOE outperforms other LoRA based models, achieving state-of-the-art (SOTA) performance. Additionally, this study explores a novel recurrent routing strategy that may inspire further advancements in enhancing the reasoning capabilities of language models., Comment: 10 pages

Recently, significant progress has been made in the study of black hole (BH) perturbations, both within the framework of general modified gravity theories and in complex environments. However, a well-established conclusion regarding the boundary conditions of the perturbed fields remains elusive. In this paper, we investigate the boundary conditions for a general perturbation at spatial infinity and the event horizon of a black hole (BH) described by a generic metric that is stationary, axisymmetric, asymptotically flat, and respects the condition of circularity. Our analysis is independent of any specific BH model or the nature of the perturbed field. In particular, by extending the formulation introduced by Teukolsky and utilizing purely geometric principles, we derive a universal expression for the boundary condition at the horizon. This expression is elegantly formulated in terms of the physical quantities at the event horizon, specifically the BH's surface gravity and angular velocity. The results presented in this work may provide valuable insights for the calculation of quasinormal modes and the gravitational waves generated by extreme-mass-ratio inspirals, extending beyond the standard Kerr case., Comment: 10 pages

The Johannsen-Psaltis (JP) metric provides a robust framework for testing the no-hair theorem of astrophysical black holes due to the regular spacetime configuration around JP black holes. Verification of this theorem through electromagnetic spectra often involves analyzing the photon sphere near black holes, which is intrinsically linked to black hole shadows. Investigating the shadow of JP black holes offers an effective approach for assessing the validity of theorem. Since the Hamilton-Jacobi equation in the JP metric does not permit exact variable separation, an approximate analytical approach is employed to calculate the shadow, while the backward ray-tracing numerical method serves as a rigorous alternative. For JP black holes with closed event horizons, the approximate analytical approach reliably reproduces results obtained through the numerical computation. However, significant discrepancies emerge for black holes with non-closed event horizons. As the deviation parameter $\epsilon_3$ increases, certain regions of the critical curve transition into non-smooth configurations. Analysis of photon trajectories in these regions reveals chaotic dynamics, which accounts for the failure of the approximate analytical method to accurately describe the shadows of JP black holes with non-closed event horizons., Comment: 37 pages, 10 figures

The diffusion transformer (DiT) architecture has attracted significant attention in image generation, achieving better fidelity, performance, and diversity. However, most existing DiT - based image generation methods focus on global - aware synthesis, and regional prompt control has been less explored. In this paper, we propose a coarse - to - fine generation pipeline for regional prompt - following generation. Specifically, we first utilize the powerful large language model (LLM) to generate both high - level descriptions of the image (such as content, topic, and objects) and low - level descriptions (such as details and style). Then, we explore the influence of cross - attention layers at different depths. We find that deeper layers are always responsible for high - level content control, while shallow layers handle low - level content control. Various prompts are injected into the proposed regional cross - attention control for coarse - to - fine generation. By using the proposed pipeline, we enhance the controllability of DiT - based image generation. Extensive quantitative and qualitative results show that our pipeline can improve the performance of the generated images., Comment: Accepted by ICASSP 2025, Github: https://github.com/ZhenXiong-dl/ICASSP2025-RCAC

Multi-modal Large Language Models (MLLMs) exhibit impressive capabilities in 2D tasks, yet encounter challenges in discerning the spatial positions, interrelations, and causal logic in scenes when transitioning from 2D to 3D representations. We find that the limitations mainly lie in: i) the high annotation cost restricting the scale-up of volumes of 3D scene data, and ii) the lack of a straightforward and effective way to perceive 3D information which results in prolonged training durations and complicates the streamlined framework. To this end, we develop pipeline based on open-source 2D MLLMs and LLMs to generate high-quality 3D-text pairs and construct 3DS-160K , to enhance the pre-training process. Leveraging this high-quality pre-training data, we introduce the 3UR-LLM model, an end-to-end 3D MLLM designed for precise interpretation of 3D scenes, showcasing exceptional capability in navigating the complexities of the physical world. 3UR-LLM directly receives 3D point cloud as input and project 3D features fused with text instructions into a manageable set of tokens. Considering the computation burden derived from these hybrid tokens, we design a 3D compressor module to cohesively compress the 3D spatial cues and textual narrative. 3UR-LLM achieves promising performance with respect to the previous SOTAs, for instance, 3UR-LLM exceeds its counterparts by 7.1\% CIDEr on ScanQA, while utilizing fewer training resources. The code and model weights for 3UR-LLM and the 3DS-160K benchmark are available at 3UR-LLM., Comment: Accepted to IEEE Transactions on Multimedia (TMM)

In this paper, we focus on the two-time correlation function (TTCF) of the cavity optomechanical system, which serves as the most popular tool in precision detection technologies. We utilize the stochastic Schrodinger equation approach to study TTCF for the cavity optomechanical system in the long-time steady state TTCF and time-dependent case. Our numerical simulations support two major conclusions: (1) long-time steady states in Markovian and non-Markovian regimes are different, resulting in the distinct TTCF, and (2) the time-dependent TTCF can reveal more information about the environment, rather than the traditional spectral function method., Comment: arXiv admin note: text overlap with arXiv:2212.13362

On top of Segment Anything Model (SAM), SAM 2 further extends its capability from image to video inputs through a memory bank mechanism and obtains a remarkable performance compared with previous methods, making it a foundation model for video segmentation task. In this paper, we aim at making SAM 2 much more efficient so that it even runs on mobile devices while maintaining a comparable performance. Despite several works optimizing SAM for better efficiency, we find they are not sufficient for SAM 2 because they all focus on compressing the image encoder, while our benchmark shows that the newly introduced memory attention blocks are also the latency bottleneck. Given this observation, we propose EdgeTAM, which leverages a novel 2D Spatial Perceiver to reduce the computational cost. In particular, the proposed 2D Spatial Perceiver encodes the densely stored frame-level memories with a lightweight Transformer that contains a fixed set of learnable queries. Given that video segmentation is a dense prediction task, we find preserving the spatial structure of the memories is essential so that the queries are split into global-level and patch-level groups. We also propose a distillation pipeline that further improves the performance without inference overhead. As a result, EdgeTAM achieves 87.7, 70.0, 72.3, and 71.7 J&F on DAVIS 2017, MOSE, SA-V val, and SA-V test, while running at 16 FPS on iPhone 15 Pro Max., Comment: Code will be released at https://github.com/facebookresearch/EdgeTAM

Large Language Models (LLMs) have demonstrated impressive capabilities in role-playing scenarios, particularly in simulating domain-specific experts using tailored prompts. This ability enables LLMs to adopt the persona of individuals with specific backgrounds, offering a cost-effective and efficient alternative to traditional, resource-intensive user studies. By mimicking human behavior, LLMs can anticipate responses based on concrete demographic or professional profiles. In this paper, we evaluate the effectiveness of LLMs in simulating individuals with diverse backgrounds and analyze the consistency of these simulated behaviors compared to real-world outcomes. In particular, we explore the potential of LLMs to interpret and respond to discharge summaries provided to patients leaving the Intensive Care Unit (ICU). We evaluate and compare with human responses the comprehensibility of discharge summaries among individuals with varying educational backgrounds, using this analysis to assess the strengths and limitations of LLM-driven simulations. Notably, when LLMs are primed with educational background information, they deliver accurate and actionable medical guidance 88% of the time. However, when other information is provided, performance significantly drops, falling below random chance levels. This preliminary study shows the potential benefits and pitfalls of automatically generating patient-specific health information from diverse populations. While LLMs show promise in simulating health personas, our results highlight critical gaps that must be addressed before they can be reliably used in clinical settings. Our findings suggest that a straightforward query-response model could outperform a more tailored approach in delivering health information. This is a crucial first step in understanding how LLMs can be optimized for personalized health communication while maintaining accuracy.

This paper introduces a cost-effective robotic handwriting system designed to replicate human-like handwriting with high precision. Combining a Raspberry Pi Pico microcontroller, 3D-printed components, and a machine learning-based handwriting generation model implemented via TensorFlow, the system converts user-supplied text into realistic stroke trajectories. By leveraging lightweight 3D-printed materials and efficient mechanical designs, the system achieves a total hardware cost of approximately \$56, significantly undercutting commercial alternatives. Experimental evaluations demonstrate handwriting precision within $\pm$0.3 millimeters and a writing speed of approximately 200 mm/min, positioning the system as a viable solution for educational, research, and assistive applications. This study seeks to lower the barriers to personalized handwriting technologies, making them accessible to a broader audience., Comment: This is an updated version of a paper originally presented at the 2024 IEEE Long Island Systems, Applications and Technology Conference (LISAT)

In this study, we analyze the observational images of a Konoplya-Zhidenko rotating non-Kerr black hole, wherein a thin accretion disk, serving as the sole background light source, is situated on the equatorial plane of the black hole. The inner boundary of the thin accretion disk extends to the event horizon, and the accretion material in the disk exhibits two different motion behaviors, that is, it moves along the critical plunging orbit inside the innermost stable circular orbit (ISCO) and follows the Keplerian orbit outside the ISCO. The shadow image is captured on the imaging plane of a zero angular momentum observer utilizing advanced fisheye camera ray-tracing techniques. The results demonstrate that an image consistently reveals a dark region encircled by a narrow photon ring, which is called the inner shadow. At low observation inclination angles, the observation intensity is highly concentrated, with the lensed image of accretion disk being superimposed on the direct image. As observation inclination angle increases, the direct and lensed images gradually separate, becoming distinctly distinguishable and forming a hat-like structure. Furthermore, variations in the parameter space and observation angle will influence pertinent image characteristics, including image symmetry, the range or deformation degree of the inner shadow. We further examined the distinctive characteristics of images observed in both prograde and retrograde accretion disk scenarios. Subsequently, we also examined the redshift distribution on the disk. The findings indicate that while variations in relevant parameters do influence the redshift distribution, the primary factor is the change in observational inclination. The observer can detect both redshift and blueshift phenomena on the screen when viewed at a higher observation angle., Comment: 32 pages, 13 figures

Neural implicit representations have recently shown promising progress in dense Simultaneous Localization And Mapping (SLAM). However, existing works have shortcomings in terms of reconstruction quality and real-time performance, mainly due to inflexible scene representation strategy without leveraging any prior information. In this paper, we introduce SP-SLAM, a novel neural RGB-D SLAM system that performs tracking and mapping in real-time. SP-SLAM computes depth images and establishes sparse voxel-encoded scene priors near the surfaces to achieve rapid convergence of the model. Subsequently, the encoding voxels computed from single-frame depth image are fused into a global volume, which facilitates high-fidelity surface reconstruction. Simultaneously, we employ tri-planes to store scene appearance information, striking a balance between achieving high-quality geometric texture mapping and minimizing memory consumption. Furthermore, in SP-SLAM, we introduce an effective optimization strategy for mapping, allowing the system to continuously optimize the poses of all historical input frames during runtime without increasing computational overhead. We conduct extensive evaluations on five benchmark datasets (Replica, ScanNet, TUM RGB-D, Synthetic RGB-D, 7-Scenes). The results demonstrate that, compared to existing methods, we achieve superior tracking accuracy and reconstruction quality, while running at a significantly faster speed.

Understanding the dynamics of photoexcited carriers is essential for advancing photoelectronic device design. Photon absorption generates electron-hole pairs, and subsequent scatterings can induce ultrafast thermalization within a picosecond, forming a quasi-equilibrium distribution with overheated electrons. The high-energy tail of this distribution enables carriers to overcome energy barriers, thereby enhancing quantum efficiency--a phenomenon known as photo-thermionic emission (PTE). Despite its importance, the onset and mechanisms of PTE remain under debate. Using real-time time-dependent density functional theory (rt-TDDFT), we investigate ultrafast carrier thermalization in two-dimensional materials graphene and PtTe2, and the results reveal distinct differences. In graphene, both electrons and holes thermalize into Fermi-Dirac distributions with good agreement to experiment, while PtTe2 exhibits anomalous high-energy tails for both electrons and holes, deviating significantly from Fermi-Dirac behavior. We attribute this anomaly to differences in orbital coupling between the two materials, from which we derive design principles for identifying optimal PTE candidates and, ultimately, improving photodetector performance., Comment: 12 pages, 5 figures, 1 table

Remote brightening (RB) is compact brightening at footpoints of magnetic loops, which are remotely-connecting to and confining an eruption in the solar atmosphere. Here, we report on observations of an RB resulting from an EUV jet with a speed of about 90\,km\,s$^{-1}$. The loops connecting the RB and the jet have an apparent length of about 59\,Mm. Intriguingly, the RB exhibits at least two episodes of brightenings, as characterised by two peaks in its lightcurve. The energies, which sustain the first and second peaks of the RB, are $6.3\times10^{26}$\,erg and $8.4\times10^{26}$\,erg, respectively, and take a significant proportion of the total energy. The first peak of the RB brightenings coincides with the jet's peak with a time delay of 12 seconds, while the second peak lags behind by 108 seconds. Besides the flows of the ejecta, we have identified two additional flows originating from the eruption site. One is relatively cool with a temperature of $log_{10}T/K=5.8-6.1$ and has a speed of about $275\pm15$\,km\,s$^{-1}$. The other is hot with a temperature of $log_{10}T/K=7.0-7.3$ and travels much faster with a speed of about 750$\pm$\,70\,km\,s$^{-1}$. We attribute the second peak of RB directly to this hot flow, which our numerical experiments suggest is the result of a slow shock wave. Considering the minimal time delay between the first peak of RB and the eruption, we infer this first episode is due to heating by nonthermal electrons. Our research demonstrates that the dynamics in an RB can offer vital insights into the nature of the corresponding eruption and help understand how the energy is distributed throughout the solar atmosphere.

Restless multi-armed bandits (RMABs) have been highly successful in optimizing sequential resource allocation across many domains. However, in many practical settings with highly scarce resources, where each agent can only receive at most one resource, such as healthcare intervention programs, the standard RMAB framework falls short. To tackle such scenarios, we introduce Finite-Horizon Single-Pull RMABs (SPRMABs), a novel variant in which each arm can only be pulled once. This single-pull constraint introduces additional complexity, rendering many existing RMAB solutions suboptimal or ineffective. %To address this, we propose using dummy states to duplicate the system, ensuring that once an arm is activated, it transitions exclusively within the dummy states. To address this shortcoming, we propose using \textit{dummy states} that expand the system and enforce the one-pull constraint. We then design a lightweight index policy for this expanded system. For the first time, we demonstrate that our index policy achieves a sub-linearly decaying average optimality gap of $\tilde{\mathcal{O}}\left(\frac{1}{\rho^{1/2}}\right)$ for a finite number of arms, where $\rho$ is the scaling factor for each arm cluster. Extensive simulations validate the proposed method, showing robust performance across various domains compared to existing benchmarks., Comment: 17 Pages, 8 figures. Accepted by AAMAS 2025

To defend against Advanced Persistent Threats on the endpoint, threat hunting employs security knowledge such as cyber threat intelligence to continuously analyze system audit logs through retrospective scanning, querying, or pattern matching, aiming to uncover attack patterns/graphs that traditional detection methods (e.g., recognition for Point of Interest) fail to capture. However, existing threat hunting systems based on provenance graphs face challenges of high false negatives, high false positives, and low efficiency when confronted with diverse attack tactics and voluminous audit logs. To address these issues, we propose a system called Actminer, which constructs query graphs from descriptive relationships in cyber threat intelligence reports for precise threat hunting (i.e., graph alignment) on provenance graphs. First, we present a heuristic search strategy based on equivalent semantic transfer to reduce false negatives. Second, we establish a filtering mechanism based on causal relationships of attack behaviors to mitigate false positives. Finally, we design a tree structure to incrementally update the alignment results, significantly improving hunting efficiency. Evaluation on the DARPA Engagement dataset demonstrates that compared to the SOTA POIROT, Actminer reduces false positives by 39.1%, eliminates all false negatives, and effectively counters adversarial attacks.

Continuous harvesting and storage of multiple fruits in a single operation allow robots to significantly reduce the travel distance required for repetitive back-and-forth movements. Traditional collision-free path planning algorithms, such as Rapidly-Exploring Random Tree (RRT) and A-star (A), often fail to meet the demands of efficient continuous fruit harvesting due to their low search efficiency and the generation of excessive redundant points. This paper presents the Interactive Local Minima Search Algorithm (ILMSA), a fast path-planning method designed for the continuous harvesting of table-top grown strawberries. The algorithm featured an interactive node expansion strategy that iteratively extended and refined collision-free path segments based on local minima points. To enable the algorithm to function in 3D, the 3D environment was projected onto multiple 2D planes, generating optimal paths on each plane. The best path was then selected, followed by integrating and smoothing the 3D path segments. Simulations demonstrated that ILMSA outperformed existing methods, reducing path length by 21.5% and planning time by 97.1% compared to 3D-RRT, while achieving 11.6% shorter paths and 25.4% fewer nodes than the Lowest Point of the Strawberry (LPS) algorithm in 3D environments. In 2D, ILMSA achieved path lengths 16.2% shorter than A, 23.4% shorter than RRT, and 20.9% shorter than RRT-Connect, while being over 96% faster and generating significantly fewer nodes. Field tests confirmed ILMSA's suitability for complex agricultural tasks, having a combined planning and execution time and an average path length that were approximately 58% and 69%, respectively, of those achieved by the LPS algorithm., Comment: Accepted by IEEE Transactions on AgriFood Electronics

Existing methods for safe multi-agent control using logic specifications like Signal Temporal Logic (STL) often face scalability issues. This is because they rely either on single-agent perspectives or on Mixed Integer Linear Programming (MILP)-based planners, which are complex to optimize. These methods have proven to be computationally expensive and inefficient when dealing with a large number of agents. To address these limitations, we present a new scalable approach to multi-agent control in this setting. Our method treats the relationships between agents using a graph structure rather than in terms of a single-agent perspective. Moreover, it combines a multi-agent collision avoidance controller with a Graph Neural Network (GNN) based planner, models the system in a decentralized fashion, and trains on STL-based objectives to generate safe and efficient plans for multiple agents, thereby optimizing the satisfaction of complex temporal specifications while also facilitating multi-agent collision avoidance. Our experiments show that our approach significantly outperforms existing methods that use a state-of-the-art MILP-based planner in terms of scalability and performance. The project website is https://jeappen.com/mastl-gcbf-website/ and the code is at https://github.com/jeappen/mastl-gcbf ., Comment: Accepted to CoRL 2024. arXiv admin note: text overlap with arXiv:2401.14554 by other authors

Domain-adaptive post-training of large language models (LLMs) has emerged as a promising approach for specialized domains such as medicine and finance. However, significant challenges remain in identifying optimal adaptation criteria and training strategies across varying data and model configurations. To address these challenges, we introduce FINDAP, a systematic and fine-grained investigation into domain-adaptive post-training of LLMs for the finance domain. Our approach begins by identifying the core capabilities required for the target domain and designing a comprehensive evaluation suite aligned with these needs. We then analyze the effectiveness of key post-training stages, including continual pretraining, instruction tuning, and preference alignment. Building on these insights, we propose an effective training recipe centered on a novel preference data distillation method, which leverages process signals from a generative reward model. The resulting model, Llama-Fin, achieves state-of-the-art performance across a wide range of financial tasks. Our analysis also highlights how each post-training stage contributes to distinct capabilities, uncovering specific challenges and effective solutions, providing valuable insights for domain adaptation of LLMs. Project page: https://github.com/SalesforceAIResearch/FinDap

In photonics, the idea of controlling light in a similar way that magnetic fields control electrons has always been attractive. It can be realized by synthesizing pseudomagnetic fields (PMFs) in photonic crystals (PhCs). Previous works mainly focus on the Landau levels and the robust transport of the chiral states. More versatile control over light using complex nonuniform PMFs such as the flexible splitting and routing of light has been elusive, which hinders their application in practical photonic integrated circuits. Here we propose an universal and systematic methodology to design nonuniform PMFs and arbitrarily control the flow of light in silicon PhCs at telecommunication wavelengths. As proofs of concept, a low-loss S-bend and a highly efficient 50:50 power splitter based on PMFs are experimentally demonstrated. A high-speed data transmission experiment is performed on these devices to prove their applicability in real communication systems. The proposed method offers a new paradigm for the exploration of fundamental physics and the development of novel nanophotonic devices.

The leaky dielectric model is widely used in simulating two-phase electrohydrodynamic (EHD) flows. One critical issue with this classical model is the assumption of Ohmic conduction, which makes it inadequate for describing the newly discovered EHD flows caused by the Onsager-Wien effect [Ryu et al., Phys. Rev. Lett. 104, 104502 (2010)]. In this paper, we proposed a phase-field lattice Boltzmann (LB) method for two-phase electrohydrodynamic flows induced by the Onsager-Wien effect. In this scheme, two LB equations are employed to resolve the incompressible Navier-Stokes equations and the conservative Allen-Cahn equation, while another three LB equations are used for solving the charge conservation equations and the electric potential equation. After we validate the developed LB method, we perform a series of numerical simulations of droplet deformation under EHD conduction phenomena. Our numerical results indicate that the presence of the Onsager-Wien effect has a significant impact on droplet deformation and charge distribution. Also, it is interesting to note that, apart from the heterocharge layers near the electrodes, a charge cloud may form between the droplet interface and the electrode in some cases. To thoroughly understand the droplet dynamics, the effects of the reference length d, the applied voltage {\Delta}{\psi}, the permittivity ratio {\epsilon}r, and the ionic mobility ratio {\mu}r on droplet deformation and charge distribution are all investigated in detail.

Recent advancements in omnimodal learning have been achieved in understanding and generation across images, text, and speech, though mainly within proprietary models. Limited omnimodal datasets and the inherent challenges associated with real-time emotional speech generation have hindered open-source progress. To address these issues, we propose openomni, a two-stage training method combining omnimodal alignment and speech generation to develop a state-of-the-art omnimodal large language model. In the alignment phase, a pre-trained speech model is further trained on text-image tasks to generalize from vision to speech in a (near) zero-shot manner, outperforming models trained on tri-modal datasets. In the speech generation phase, a lightweight decoder facilitates real-time emotional speech through training on speech tasks and preference learning. Experiments demonstrate that openomni consistently improves across omnimodal, vision-language, and speech-language evaluations, enabling natural, emotion-rich dialogues and real-time emotional speech generation.